[Build/Upgrade Log] The Optiplex 755 SFF twins. 2 identical machines, 2 completely different purposes. - Overclock.net - An Overclocking Community

Forum Jump: 

[Build/Upgrade Log] The Optiplex 755 SFF twins. 2 identical machines, 2 completely different purposes.

Reply
 
Thread Tools
post #1 of 29 (permalink) Old 04-26-2014, 08:14 PM - Thread Starter
New to Overclock.net
 
Join Date: Feb 2007
Location: Chicago
Posts: 973
Rep: 26 (Unique: 24)


At first, I wasn't sure whether to post this, but realized there might be some people interested due to its "off-the-beaten-path" nature. Then I wasn't sure whether to call it a build log since I'm not starting from scratch, but working from an existing skeleton and changing everything. So I suppose a build/upgrade log was more accurate. As a disclaimer, it definitely won't be an insane build like the one in my signature, however I think it'll raise some eyebrows at the very least. I call them twins because they both have the same ship date (per their service tags on Dell's support site) of April 18, 2008.


Backstory
When I realized that the Windows XP support cutoff date had come and gone, I immediately starting looking around for viable options to replace my mother's computer because A) I needed to get her off of Windows XP and B) it's not exactly a quick machine anymore. At first I was thinking laptop so she had more mobility and could carry it around the house, but at the price point I was looking at of $150-200, the available laptops were god awful. All the better when she mentioned that she hates laptops. So I began looking at desktops, and sure enough there was way more bang-for-the-buck to be had. In particular, there was one Craigslist listing that caught my eye for a Dell Optiplex 755 small form factor. I started to do my research, and it turned out these were INCREDIBLY popular business machines. Heck, if you start typing into Google "Dell Optiplex", the 755 is the very first result. I did some price comparing and the seller had it for a really good price. I met with him, tested the machine, opened it up to find a very clean interior, and paid him for it. It was originally a Core 2 Duo E6850 with 1GB of DDR2-667 (unknown latency) RAM for $80, and he had the option that for every 1GB more I wanted it would be $10 more. I ended up swapping out the two 512MB sticks for two 2GB sticks of Hynix DDR2-800 6-6-6-18 1.8V RAM. Even at what would have been $110 (original $80 + $30 for the extra 3GB of RAM) I think it would have been a good deal, but I walked out with it for $95 which is a steal IMO.

The problem started when I researched it some more and realized that it supported a HUGE array of CPUs, all the way from a lowly Core 2 Duo "Allendale" E4300 to a powerful Core 2 Quad "Yorkfield" Q9650. Then the gears in my head really starting turning when I thought SSD and especially GPU. Thanks go out to Galaxy when I discovered they are the only company that has made a low-profile version within the current 700-series generation of Nvidia (the credit goes to OCN user AlphaC who brought this to my attention). If it wasn't for them, I would have been stuck putting in a Zotac card from the previous generation or even worse a Radeon HD 7750 (worse because I hate ATI). In fact, I give a very special thanks (and also hate for enabling this crazy project to take root with me tongue.gif ) goes out to Dave from Galaxy. I was extremely lucky to find out that Galaxy's RMA/returns office in Volo, Illinois was a mere 45 minute drive from my place. He was really helpful and great to allow me to test fit the card first and hook up one of their test monitors to get it booted into BIOS, which it did flawlessly. It's at that point he said, "This is is crazy. You realize the insanity of having this card in here and actually WORKING? It makes for THE perfect little LAN box." That pretty much is what sealed the decision, which is that I wanted to keep the machine for myself. The whole point of this endeavor was not to own yet another machine, but to get a new machine for mom. So I actually contacted the seller to ask if he had anymore, and he did. I ended up getting another clean example with a Core 2 Duo E4500 and 2GB (four 512MB sticks) of Samsung and Hynix DDR2-667 5-5-5-15 1.8V RAM for $65. I purposely got the cheapest thing possible as everything will be getting upgraded. I don't have the parts for the 1st one yet (which is the one that I'll be giving to mom once I finish it as part of her Mother's Day gift. Remember fellas, like Mr. T says you gotta treat your mother right thumb.gif ). However, I'm sure everyone is WAY more interested in the 2nd one that is to be reborn as a portable powerhouse. So let's get to the parts.

For the SSD I went with a Seagate 600 Pro 240GB that I picked up from TigerDirect at $120 out-the-door thanks to a sale going on along with a $10 coupon that was specifically for TigerDirect from RetailMeNot.


I got exceptionally lucky with the RAM. Not only did I find 4 brand new 2GB sticks of G.Skill DDR2-800 5-5-5-15 1.8V RAM for $75 shipped, but they were from the same production (hard to see on the hologram but it says 2014-April) and are SEQUENTIAL. It only really matters that they would be paired for running in dual channel mode, but this was pretty cool to find 4 sticks produced back-to-back.


The best CPU for Socket 775 from the Core 2 family era: the Core 2 Quad Q9550S. But wait you might say, isn't there a Q9650 that runs at a slightly higher clock speed per core? You would be correct, however I specifically selected the Q9550S because it's the fastest Core 2 Quad they made with a TDP of 65W compared to the Q9650 at 95W. I'l gladly lose a mere 5% in clock speed in order to drop heat output and more importantly power usage by 30%, because I want all the extra headroom I can get when taking the GPU into account. I got this for a good deal of $125 to my door.


And finally, the centerpiece: the Galaxy GTX 750 Ti GC Slim. Once again, props go to OCN user AlphaC for telling me this thing existed and to Dave at Galaxy for helping me confirm it would work. Picked it up for $160 total with the low-profile brackets.





All of the above is to go inside here. How you ask? Well, you'll just have to wait and see. wink.gif


In order to give this log a little more meaning, I'll be doing some baseline benchmarking and data collection as a reference for myself and for others who may come across this thread in a Google search for Dell Optiplex 755. Of course, this will be compared to the final numbers once both machines are finished to show all the changes. I'll try to include things like power usage while idling at desktop, full load, and completely shutdown, max temps of the CPU and GPU, benchmarks for the CPU and GPU, and overall system performance which brings the SSD into the equation. To that end, for the GPU I'll use 3DMark, 3DMark 11, 3DMark Vantage, along with Heaven and Valley from Uniengine. I'm not sure what to use for the CPU as SuperPi doesn't have a multi-core version, and I don't know what's the current go-to for CPU performance benchmarking. As for testing the system as a whole, I would use SiSoftware Sandra unless there is a better choice to use nowadays.

So if anyone can point me to CPU specific and overall system tools, I would greatly appreciate it.

To sum up, here's what's going down for machine #2 (machine #1 will likely occur after as I'm waiting on parts.)
  1. CPU - Intel Core 2 Duo E4500 --> Intel Core 2 Quad Q9550s
  2. RAM - 4x512MB mix of Samsung and Hynix DDR2-667 5-5-5-15 1.8V --> 4x2GB G.Skill DDR2-800 5-5-5-15 1.8V
  3. Storage - Western Digital Caviar SE WD800JD 80GB HDD --> Seagate 600 Pro 240GB SSD
  4. GPU - Integrated Intel GMA 3100 --> Galaxy NVIDIA GTX 750 Ti GC Slim 2GB GDDR5

If you have gotten this far, I thank you for reading through the thread. As you can see in my other build, I tend to write lengthy posts but only because I like to be thorough. So if you're interested to see how everything goes, go ahead and subscribe to the thread. I'll try to make regular updates on the weekends if not sooner (depending on if I have time during the week with work) with pictures and/or numbers.

First thing's first after I finish this post will be to run the baseline numbers. I've got my Kill-A-Watt setup and ready to go, along with Prime 95, CoreTemp, HWMonitor, and RealTemp. Then I'll run the battery of tests in SiSoftware Sandra. I'm not going to bother with any GPU tests as the GMA 3100 is an extremely bad integrated graphics solution, and I don't think any benchmarking tool would even run on it.

Stay tuned! I'm excited to once again be doing this, and to see what the final results will be. biggrin.gif

UPDATE 5/3/14
Apologies for posting this today rather than last Sunday, but all the testing I wanted to end up doing took way more time than I initially thought it would. Not to mention some things came up that ate up quite a bit of time. Now that the weekend is here, I can get back to work on the machines. In any event, I have completed all the testing and obtained all the baseline numbers, information, and benchmarks on machine #2 which is what all that good stuff in the first post is getting installed in. So let's get started smile.gif

First up will be verification of the current components.

BIOS report on CPU Results (Click to show)

BIOS report on RAM Results (Click to show)

CPU-Z report Results (Click to show)

GPU-Z report. I thought it was interesting that it's not reporting it as GMA 3100, which is what it actually is. Not sure why that is. Results (Click to show)

Speccy report Results (Click to show)



Next, I used my P3 Kill A Watt to see what kind of wattage the system is using in various states.
Results (Click to show)
Powered off. Fantastic to see it have essentially zero vampire power draw.


Sitting idle at the desktop


Highest load possible using IntelBurnTest. Interesting note here is the highest it would show using Prime95 on the In-place large FFTs torture test was 93 watts.


Summary of power usage: 1 watt while powered off, 48 watts while idle at desktop, 100 watts with full CPU load via IntelBurnTest.

Next, I timed the speed of the system's boot up and shutdown. Results (Click to show)
Cold boot (time from pressing the button while powered off to sitting idle at the desktop). Ignore the 2nd time as I accidentally hit the lap button instead of stop.


Restart (time from hitting enter on restart to sitting idle at the desktop).


Shutdown (time from hitting enter on shutdown to power off state).


Summary of times: 61.1 seconds for cold boot, 65.4 seconds for restart, 8.5 seconds for shutdown

Now let's move on to benchmark numbers. For the CPU, I used Cinebench, IntelBurnTest, and wPrime along with the tests that were part of the overall system test of Passmark PerformanceTest and SiSoftware Sandra. For the RAM, I'll refer to the numbers given by PerformanceTest and Sandra. For the storage drive, I used CrystalDiskMark and I'll refer to the numbers given by PerformanceTest and Sandra. For the GPU, I didn't bother testing as well all know how bad integrated graphics are, not to mention none of the tests would work. Finally, overall system performance was tested with PCMark 07, PCMark 08, PerformanceTest, Sandra, and the Windows Performance Index.

Cinebench Results (Click to show)

IntelBurnTest. I also checked to see what was the max temp the CPU was getting to. I've read that Core Temp is THE program to use for checking CPU temps due to it reading the diode within the CPU for the highest accuracy. I'm not sure why RealTemp was consistently showing a temp of 5-6 degrees C higher than Core Temp. Results (Click to show)

wPrime. Settings were the 32M test with thread count set to 2. Results (Click to show)

CrystalDiskMark. Settings were Test Data set to Default (Random) with 3 runs for each test. I'll paste the text output after the picture. I definitely had to laugh at these abysmal numbers. Results (Click to show)

CrystalDiskMark 3.0.3 x64 (C) 2007-2013 hiyohiyo
Crystal Dew World : http://crystalmark.info/
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

Sequential Read : 55.515 MB/s
Sequential Write : 53.892 MB/s
Random Read 512KB : 23.067 MB/s
Random Write 512KB : 27.693 MB/s
Random Read 4KB (QD=1) : 0.313 MB/s [ 76.4 IOPS]
Random Write 4KB (QD=1) : 0.858 MB/s [ 209.6 IOPS]
Random Read 4KB (QD=32) : 0.308 MB/s [ 75.1 IOPS]
Random Write 4KB (QD=32) : 0.865 MB/s [ 211.3 IOPS]

Test : 1000 MB [C: 21.6% (16.1/74.4 GB)] (x3)
Date : 2014/04/27 16:28:39
OS : Windows 7 Professional SP1 [6.1 Build 7601] (x64)

PCMark 07. I did 3 runs and got 1095, 1102, and 1101. I kept the results of the best run, which is linked below. Not sure why it says GMA X3500 instead of GMA 3100, but I guess Futuremark took pity on seeing a GMA 3100 being tested and bumped it up haha. Results (Click to show)
PCMark 08. The GMA 3100 is so bad that I couldn't even complete either the Home 3.0 or Work 2.0 tests in either Conventional or Accelerated Mode, and it wouldn't let the Creative test run because it said "Your hardware is not compatible with this test. To run this test you need a DirectX 11 compatible GPU." I was able to able to get some scores though. Results (Click to show)
Home 3.0 tests. In both instances it would crash on Test 4 Casual Gaming.
http://www.3dmark.com/compare/pcm8hm3/21014/pcm8hm3/21012

Work 2.0 tests. In both instances it would crash on Test 5 Video Chat. No link because I accidentally deleted the Conventional test result.


This was the error I would get every time it failed on Test 4 of the Home test.

This was the error I would get every time it failed on Test 5 of the Work test.

SiSoftware Sandra. I really don't like the way results are presented in this program, and there was no easy way to take screens. So instead I'll cut and paste the relevant info from the various text outputs. Results (Click to show)
Processor Arithmetic Results (Click to show)
Benchmark Results
Aggregate Native Performance : 13.68GOPS
Dhrystone Integer Native ALU : 18.72GIPS
Whetstone Single-float Native SSE3 : 13GFLOPS
Whetstone Double-float Native SSE3 : 7.69GFLOPS
Whetstone Aggregated-float Native : 10GFLOPS
Performance per Thread
Aggregate Native Performance : 6.84GOPS
Dhrystone Integer Native ALU : 9.36GIPS
Whetstone Single-float Native SSE3 : 6.48GFLOPS
Whetstone Double-float Native SSE3 : 3.85GFLOPS
Whetstone Aggregated-float Native : 5GFLOPS
No. Threads : 2
Performance vs. Speed
Aggregate Native Performance : 6.23MOPS/MHz
Dhrystone Integer Native ALU : 8.53MIPS/MHz
Whetstone Single-float Native SSE3 : 5.91MFLOPS/MHz
Whetstone Double-float Native SSE3 : 3.50MFLOPS/MHz
Whetstone Aggregated-float Native : 4.55MFLOPS/MHz

Processor Multi-Media Results (Click to show)
Benchmark Results
Aggregate Multi-Media Native Performance : 25.18MPix/s
Multi-Media Integer Native x8 SSSE3 : 30.34MPix/s
Multi-Media Long-int Native x1 ALU : 6.19MPix/s
Multi-Media Quad-int Native x1 ALU : 316kPix/s
Multi-Media Single-float Native x8 SSE2 : 31.14MPix/s
Multi-Media Double-float Native x4 SSE2 : 16.9MPix/s
Multi-Media Quad-float Native x1 SSE2 : 728kPix/s
Aggregate Multi-Media Native Performance
Multi-Media Aggregate-int Native : 13.7MPix/s
Multi-Media Aggregate-float Native : 23MPix/s
Performance per Thread
Aggregate Multi-Media Native Performance : 12.59MPix/s
Multi-Media Integer Native x8 SSSE3 : 15.17MPix/s
Multi-Media Long-int Native x1 ALU : 3.1MPix/s
Multi-Media Quad-int Native x1 ALU : 158kPix/s
Multi-Media Single-float Native x8 SSE2 : 15.57MPix/s
Multi-Media Double-float Native x4 SSE2 : 8.45MPix/s
Multi-Media Quad-float Native x1 SSE2 : 364kPix/s
No. Threads : 2
Performance vs. Speed
Aggregate Multi-Media Native Performance : 11.47kPix/s/MHz
Multi-Media Integer Native x8 SSSE3 : 13.82kPix/s/MHz
Multi-Media Long-int Native x1 ALU : 2.82kPix/s/MHz
Multi-Media Quad-int Native x1 ALU : 0.14kPix/s/MHz
Multi-Media Single-float Native x8 SSE2 : 14.19kPix/s/MHz
Multi-Media Double-float Native x4 SSE2 : 7.70kPix/s/MHz
Multi-Media Quad-float Native x1 SSE2 : 0.33kPix/s/MHz

Processor Multi-Core Efficiency Results (Click to show)
Benchmark Results
Inter-Core Bandwidth : 2.27GB/s
Benchmark Results
Inter-Core Latency : 55.6ns
Performance per Thread
Inter-Core Bandwidth : 1.14GB/s
No. Threads : 2
Performance vs. Speed
Inter-Core Bandwidth : 1.06MB/s/MHz
Inter-Core Latency : 0.25ns/MHz

Memory Bandwidth Results (Click to show)
Benchmark Results
Aggregate Memory Performance : 3.7GB/s
Integer Memory Bandwidth B/F SSE2/128 : 3.71GB/s
Float Memory Bandwidth B/F SSE2/128 : 3.7GB/s
Benchmark Timings
Time to Copy Capacity : 540µs
Performance per Thread
Aggregate Memory Performance : 1.85GB/s
Integer Memory Bandwidth B/F SSE2/128 : 3.71GB/s
Float Memory Bandwidth B/F SSE2/128 : 3.7GB/s
No. Threads : 2
Performance vs. Power
Chipset(s)/Memory Power : 18.100W
Aggregate Memory Performance : 209.56MB/s/W
Integer Memory Bandwidth B/F SSE2/128 : 209.72MB/s/W
Float Memory Bandwidth B/F SSE2/128 : 209.39MB/s/W
Capacity vs. Power
Memory Capacity : 113MB/W
Performance vs. Speed
Aggregate Memory Performance : 5.70MB/s/MHz
Integer Memory Bandwidth B/F SSE2/128 : 5.70MB/s/MHz
Float Memory Bandwidth B/F SSE2/128 : 5.69MB/s/MHz
Integer Memory Bandwidth
Assignment : 3.64GB/s
Scaling : 3.6GB/s
Addition : 3.81GB/s
Triad : 3.77GB/s
Data Item Size : 16bytes
Buffering Used : Yes
Offset Displacement : Yes
Bandwidth Efficiency : 57.61%
Float Memory Bandwidth
Assignment : 3.62GB/s
Scaling : 3.6GB/s
Addition : 3.81GB/s
Triad : 3.77GB/s
Data Item Size : 16bytes
Buffering Used : Yes
Offset Displacement : Yes
Bandwidth Efficiency : 57.72%

Cache Bandwidth Results (Click to show)
Benchmark Results
Cache/Memory Bandwidth SSE2/128 : 17.63GB/s
L1D (1st Level) Data Cache : 127.31GB/s
L2 (2nd Level) Data/Unified Cache : 36.65GB/s
Benchmark Results
Speed Factor : 48.00
Performance per Thread
Cache/Memory Bandwidth : 8.81GB/s
L1D (1st Level) Data Cache : 63.65GB/s
L2 (2nd Level) Data/Unified Cache : 18.33GB/s
No. Threads : 2
Performance vs. Power
Processor(s)/Chipset(s)/Memory Power : 18.100W
Cache/Memory Bandwidth SSE2/128 : 997.24MB/s/W
Capacity vs. Power
Total Cache Size : 116.69kB/W
Performance vs. Speed
Cache/Memory Bandwidth SSE2/128 : 8.22MB/s/MHz
Cache/Memory Results Breakdown
Data Item Size : 16bytes
Buffering Used : No
Offset Displacement : Yes
Detailed Results
2kB Data Set : 121.58GB/s (52.2GB/s - 121.58GB/s)
4kB Data Set : 123.66GB/s (47.41GB/s - 123.66GB/s)
8kB Data Set : 131.76GB/s (50.75GB/s - 131.76GB/s)
16kB Data Set : 133.37GB/s (58.53GB/s - 133.37GB/s)
32kB Data Set : 126.17GB/s (45.92GB/s - 126.17GB/s)
64kB Data Set : 99.88GB/s (40.46GB/s - 99.88GB/s)
128kB Data Set : 38.79GB/s (26.13GB/s - 38.79GB/s)
256kB Data Set : 36.77GB/s (25.3GB/s - 36.77GB/s)
512kB Data Set : 36.32GB/s (24.68GB/s - 36.32GB/s)
1MB Data Set : 34.72GB/s (23GB/s - 34.72GB/s)
2MB Data Set : 28GB/s (19GB/s - 28GB/s)
4MB Data Set : 6.31GB/s (4.32GB/s - 6.31GB/s)
8MB Data Set : 3.7GB/s (3.55GB/s - 3.7GB/s)
16MB Data Set : 3.69GB/s (3.52GB/s - 3.69GB/s)
64MB Data Set : 3.5GB/s (3.33GB/s - 3.5GB/s)
256MB Data Set : 2.78GB/s (2.65GB/s - 2.78GB/s)

Cache and Memory Latency Results (Click to show)
Benchmark Results
Memory Latency : 55.0ns
Decimal Numeral System (base 10) : 1s = 1000ms, 1ms = 1000µs, 1µs = 1000ns, etc.
L1D (1st Level) Data Cache : 3.0clocks
L2 (2nd Level) Data/Unified Cache : 14.2clocks
Speed Factor : 40.30
Performance vs. Power
Chipset(s)/Memory Power : 18.100W
Memory Latency : 3.04ns/W
Capacity vs. Power
Memory Capacity : 113.15MB/W
Performance vs. Speed
Memory Latency : 0.08ns/MHz
Detailed Results
2kB Range : 1.4ns / 3.0clocks
4kB Range : 1.4ns / 3.0clocks
8kB Range : 1.4ns / 3.0clocks
16kB Range : 1.4ns / 3.0clocks
32kB Range : 1.4ns / 3.0clocks
64kB Range : 6.4ns / 14.0clocks
128kB Range : 6.5ns / 14.0clocks
256kB Range : 6.5ns / 14.0clocks
512kB Range : 6.5ns / 14.0clocks
1MB Range : 6.6ns / 15.0clocks
2MB Range : 10.8ns / 24.0clocks
4MB Range : 55.0ns / 121.0clocks
8MB Range : 55.0ns / 121.0clocks
16MB Range : 55.0ns / 121.0clocks
32MB Range : 55.0ns / 121.0clocks
64MB Range : 55.0ns / 121.0clocks
256MB Range : 55.0ns / 121.0clocks

File System Bandwidth Results (Click to show)
Benchmark Results
Drive Score : 42.46MB/s
Sequential Read Bandwidth : 52.15MB/s
Sequential Write Bandwidth : 49.72MB/s
Benchmark Results
Random Access Time : 15.63ms
Benchmark Timings
Time to Read Capacity : 25 minute(s), 34 second(s)
Time to Write Capacity : 26 minute(s), 49 second(s)
Performance per Thread
Drive Score : 21.23MB/s
Sequential Read Bandwidth : 26.07MB/s
Sequential Write Bandwidth : 24.86MB/s
No. Threads : 2
Performance vs. Speed
Drive Score : 6.04kB/s/rpm
Sequential Read Bandwidth : 7.42kB/s/rpm
Sequential Write Bandwidth : 7.07kB/s/rpm
Random Access Time : 2.171µs/rpm
Detailed Results
Buffered Read Bandwidth : 196.14MB/s
Sequential Read Bandwidth : 52.15MB/s
Random Read Bandwidth : 28.73MB/s
Buffered Write Bandwidth : 181.93MB/s
Sequential Write Bandwidth : 49.72MB/s
Random Write Bandwidth : 29.68MB/s
Random Access Time : 15.63ms

Physical Disk Read Results (Click to show)
Benchmark Results
Drive Score : 49.69MB/s
Benchmark Results
Random Access Time : 19.73ms
Benchmark Timings
Time to Read Capacity : 26 minute(s), 50 second(s)
Performance per Thread
Drive Score : 24.85MB/s
No. Threads : 2
Performance vs. Speed
Drive Score : 7.07kB/s/rpm
Random Access Time : 2.741µs/rpm
Detailed Results
Speed at position 0% : 57.89MB/s (25.25MB/s - 57.89MB/s) (117%)
Speed at position 3% : 57.62MB/s (22.7MB/s - 57.62MB/s) (116%)
Speed at position 7% : 57.53MB/s (21.38MB/s - 57.53MB/s) (116%)
Speed at position 10% : 57.38MB/s (22.6MB/s - 57.38MB/s) (115%)
Speed at position 13% : 57.18MB/s (20.75MB/s - 57.18MB/s) (115%)
Speed at position 17% : 57.3MB/s (21.81MB/s - 57.3MB/s) (115%)
Speed at position 20% : 56.31MB/s (20MB/s - 56.31MB/s) (113%)
Speed at position 23% : 56.66MB/s (23.87MB/s - 56.66MB/s) (114%)
Speed at position 27% : 55.91MB/s (24.1MB/s - 55.91MB/s) (113%)
Speed at position 30% : 55.1MB/s (23.25MB/s - 55.1MB/s) (111%)
Speed at position 33% : 54.91MB/s (20.4MB/s - 54.91MB/s) (110%)
Speed at position 37% : 53.91MB/s (20.4MB/s - 53.91MB/s) (108%)
Speed at position 40% : 54.48MB/s (21.82MB/s - 54.48MB/s) (110%)
Speed at position 43% : 53MB/s (21.24MB/s - 53MB/s) (107%)
Speed at position 47% : 53.17MB/s (22.83MB/s - 53.17MB/s) (107%)
Speed at position 50% : 51.69MB/s (21.42MB/s - 51.69MB/s) (104%)
Speed at position 53% : 51.51MB/s (19.61MB/s - 51.51MB/s) (104%)
Speed at position 57% : 50.42MB/s (21.83MB/s - 50.42MB/s) (101%)
Speed at position 60% : 50.25MB/s (20MB/s - 50.25MB/s) (101%)
Speed at position 63% : 47.39MB/s (18MB/s - 47.39MB/s) (95%)
Speed at position 67% : 47.47MB/s (18.72MB/s - 47.47MB/s) (96%)
Speed at position 70% : 46.15MB/s (19.47MB/s - 46.15MB/s) (93%)
Speed at position 73% : 44.62MB/s (17.88MB/s - 44.62MB/s) (90%)
Speed at position 77% : 43.27MB/s (17.1MB/s - 43.27MB/s) (87%)
Speed at position 80% : 43.43MB/s (18.87MB/s - 43.43MB/s) (87%)
Speed at position 83% : 41.92MB/s (17.36MB/s - 41.92MB/s) (84%)
Speed at position 87% : 40MB/s (17.86MB/s - 40MB/s) (80%)
Speed at position 90% : 37.72MB/s (15.32MB/s - 37.72MB/s) (76%)
Speed at position 93% : 37.77MB/s (16.13MB/s - 37.77MB/s) (76%)
Speed at position 97% : 35MB/s (15.33MB/s - 35MB/s) (70%)
Speed at position 100% : 33.51MB/s (13.54MB/s - 33.51MB/s) (67%)
Random Access Time : 19.73ms (13.6ms - 65.36ms)
Full Stroke Access Time : 19.8ms (13ms - 66.24ms)

System Score Results (Click to show)
Processor Arithmetic
Aggregated Score : 13.86GOPS
Processor Multi-Media
Aggregated Score : 23.96MPix/s
Cryptography
Aggregated Score : 0.180GB/s
Processor Financial Analysis
Aggregated Score : 1.50kOPT/s
.NET Arithmetic
Aggregated Score : 6.48GOPS
.NET Multi-Media
Aggregated Score : 3.69MPix/s
Memory Bandwidth
Aggregated Score : 3.749GB/s
Cache & Memory Latency
Aggregated Score : 57.4ns
File System Bandwidth
Aggregated Score : 43.736MB/s
File System I/O
Aggregated Score : 881.8IOPS
Overall Score
Aggregated Score : 0.86kPT


PassMark PerformanceTest. This was by far my favorite of the system benchmarking tools. So much better than Sandra IMHO. Results (Click to show)
Settings


PassMark Rating


Summary. Link to PassMark's site for my baseline - http://www.passmark.com/baselines/V8/display.php?id=23397800793


CPU Mark


2D Graphics Mark


3D Graphics Mark - Normally there is also a score for Directx 10, Directx 11, and Directcompute, but as stated before the GMA 3100 is extremely bad.


Memory Mark


Disk Mark

Windows Experience Index. Not sure what to make of that caution message at the bottom. Results (Click to show)


Summary of all benchmarks Results (Click to show)
  • Cinebench - CPU 105 cb
  • IntelBurnTest (Best of 3 runs) - 142.442 seconds and 10.9713 GFLOPS
  • wPrime 32M test (Best of 3 runs) - 42.001 seconds
  • CrystalDiskMark with Default Random test date (Best of 3 runs) - Sequential Read 55.515 MB/s, Sequential Write 53.892 MB/s, Random 4KB read @ QD1 of 0.313 MB/s (76.4 IOPS), Random 4KB write @ QD1 of 0.858 MB/s (209.6 IOPS), Random 4KB read @ QD32 of 0.308 MB/s (75.1 IOPS), Random 4KB write @ QD32 of 0.865 MB/s (211.3 IOPS)
  • PCMark 07 (Best of 3 runs) - 1102
  • PCMark 08 (Best of 3 runs) - No score due to graphics test failing on both Home 3.0 and Work 2.0 tests
  • SiSoftware Sandra Overall Aggregated Score - 0.86kPT
  • PassMark PerformanceTest - PassMark rating of 142.9 with subscores of CPU Mark 1583, 2D Graphics Mark 294.5, 3D Graphics Mark 13.3, Memory Mark 630, and Disk Mark 376.3
  • Windows Experience Index - Base Score of 3.3 with subscores of Processor 5.4, Memory (RAM) 5.4, Graphics 3.8, Gaming graphics 3.3, and Primary hard disk 5.4


PHEW! That took forever to type out and properly format, and I hope it is easy to read and everything makes sense. It's definitely a lot of numbers and data, but I wanted to cover everything and be as thorough as possible.

What I'll be working on next is doing a less in-depth version of this for machine #1 aka Mom's Machine because the CPU arrived and I want to get the shortened version of the above done today. This way I can install the CPU, GPU, and SSD to get it ready for next Sunday the 11th.

UPDATE 5/5/14
I finished collection of all the data, numbers, and benchmarks for the baseline for machine #1. This is what's currently installed in it (won't list the GPU as it's the same integrated one as on machine #2):
  1. CPU - Intel Core 2 Duo E6850
  2. RAM - 2x2GB Hynix DDR2-800 6-6-6-18 1.8V
  3. Storage - Seagate Barracuda 7200.10 ST380815AS 80GB HDD

As you can see, it's virtually identical to machine #2 minus the (admittedly) much better CPU. Therefore, you'd think the benchmarks would be the same except the CPU-focused ones. I thought the same thing, and turns out we're both wrong. Let's go through the results. I'll follow the same format as in the previous update, with the exception of including machine #2's results as a comparison where applicable.

Verification of the current components. Results (Click to show)
BIOS report on CPU


BIOS report on RAM


CPU-Z report

Wattage use.
Results (Click to show)
Powered off. The same as machine #2 (unsurprisingly).


Sitting idle at the desktop. Surprised it's only 3 watts more even though the E6850 idles at 2.0 GHz compared to the E4500's 1.2 GHz.


Highest load possible using IntelBurnTest. Once again, IBT produced a higher result. And while I did catch it spike to 132 watts for a few seconds once or twice, this is where it consistently peaked.


Summary of power usage: 1 watt while powered off, 51 watts (vs. 48 watts) while idle at desktop, 129 watts (vs. 100 watts) with full CPU load via IntelBurnTest.

Speed of the system's boot up and shutdown.
Results (Click to show)
Cold boot


Restart


Shutdown


Summary of times: 60.7 seconds (vs. 61.1 seconds) for cold boot, 62.3 seconds (vs. 65.4 seconds) for restart, 8.9 seconds (vs 8.5 seconds) for shutdown.

Benchmark numbers. Same format and software used as before. This time rather than putting the comparison after each test, I'll combine everything at the end in the summary section. That way it's a good reference point, and all the info is together on one screen.

Cinebench Results (Click to show)

IntelBurnTest. Considering the large increase in clock speed with commensurate power draw increase, the higher max temp is not surprising. And RealTemp once again was showing a temp of 5-6 degrees C higher than Core Temp. Also, what I did this time was to run the test at 1462MB so the results could be directly compared to machine #2 (which was how much it used when set to stress level maximum due to it's limited 2GB of RAM). Then I ran a second test at 3GB (3072MB) so that I can later compare it to machine #2 once machine #2 has the larger RAM amount of 8GB.
Results (Click to show)
1462MB


3072MB

wPrime Results (Click to show)

CrystalDiskMark Results (Click to show)

CrystalDiskMark 3.0.3 x64 (C) 2007-2013 hiyohiyo
Crystal Dew World : http://crystalmark.info/
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

Sequential Read : 63.289 MB/s
Sequential Write : 62.419 MB/s
Random Read 512KB : 30.700 MB/s
Random Write 512KB : 26.350 MB/s
Random Read 4KB (QD=1) : 0.464 MB/s [ 113.3 IOPS]
Random Write 4KB (QD=1) : 0.906 MB/s [ 221.1 IOPS]
Random Read 4KB (QD=32) : 0.834 MB/s [ 203.6 IOPS]
Random Write 4KB (QD=32) : 0.946 MB/s [ 231.0 IOPS]

Test : 1000 MB [C: 34.3% (25.5/74.4 GB)] (x3)
Date : 2014/05/04 0:44:01
OS : Windows 7 Professional SP1 [6.1 Build 7601] (x64)

PCMark 07 Results (Click to show)

PCMark 08. Same issues as before. Results (Click to show)
SiSoftware Sandra. Cutting down this section to just the overall system score. This time the pie chart is actually useful. Blue is this one, and green is machine #2. Results (Click to show)

System Score
Processor Arithmetic
Aggregated Score : 19.45GOPS
Processor Multi-Media
Aggregated Score : 33.94MPix/s
Cryptography
Aggregated Score : 0.249GB/s
Processor Financial Analysis
Aggregated Score : 2.10kOPT/s
.NET Arithmetic
Aggregated Score : 8.76GOPS
.NET Multi-Media
Aggregated Score : 5.21MPix/s
Memory Bandwidth
Aggregated Score : 5.559GB/s
Cache & Memory Latency
Aggregated Score : 47.2ns
File System Bandwidth
Aggregated Score : 50.321MB/s
File System I/O
Aggregated Score : 1582.9IOPS
Overall Score
Aggregated Score : 1.20kPT


PassMark PerformanceTest. Settings were same as before except I changed the iterations from 5 to 3, because 5 takes way too long. 3 should still produce a good average though. I like PassMark even more now as it shows its colors (pun intended) as a fantastic comparison tool. In this instance, green corresponds to this one and red to machine #2. Results (Click to show)
PassMark Rating


Summary. Link to PassMark's site for my baseline - http://www.passmark.com/baselines/V8/display.php?id=23472417247


CPU Mark


2D Graphics Mark


3D Graphics Mark


Memory Mark


Disk Mark


Windows Experience Index Results (Click to show)

Summary of all benchmarks Results (Click to show)
  • Cinebench - CPU 149cb (vs. 105 cb). ~42% increase.
  • IntelBurnTest (Best of 3 runs) - 78.313 seconds (vs. 142.442 seconds) and 19.9555 GFLOPS (vs. 10.9713 GFLOPS). ~81% faster and ~81% increase, respectively.
  • wPrime 32M test (Best of 3 runs) - 30.623 seconds (vs. 42.001 seconds). ~38% faster.
  • CrystalDiskMark with Default Random test date (Best of 3 runs) - Sequential Read 63.289 MB/s (vs. 55.515 MB/s), Sequential Write 62.419 MB/s (vs. 53.892 MB/s), Random 4KB read @ QD1 of 0.464 MB/s [113.3 IOPS] (vs. 0.313 MB/s [76.4 IOPS]), Random 4KB write @ QD1 of 0.906 MB/s [221.1 IOPS] (vs. 0.858 MB/s [209.6 IOPS]), Random 4KB read @ QD32 of 0.834 MB/s [203.6 IOPS] (vs. 0.308 MB/s [75.1 IOPS]), Random 4KB write @ QD32 of 0.946 MB/s [231.0 IOPS] (vs. 0.865 MB/s [211.3 IOPS]). ~14% faster on sequential, 5-48% faster on random 4KB QD1, 10-170% faster on random 4KB QD32.
  • PCMark 07 (Best of 3 runs) - 1366 (vs. 1102). ~24% increase.
  • PCMark 08 (Best of 3 runs) - No score due to graphics test failing on both Home 3.0 and Work 2.0 tests
  • SiSoftware Sandra Overall Aggregated Score - 1.20kPT (vs. 0.86kPT). ~40% increase.
  • PassMark PerformanceTest - PassMark rating of 191.6 (vs. 142.9) with subscores of CPU Mark 2298 (vs. 1583), 2D Graphics Mark 438.8 (vs. 294.5), 3D Graphics Mark 17.8 (vs. 13.3), Memory Mark 888 (vs. 630), and Disk Mark 424.2 (vs. 376.3). ~34%, ~45%, ~49%, ~33%, ~41%, and ~13% increase, respectively.
  • Windows Experience Index - Base Score of 3.4 (vs. 3.3) with subscores of Processor 6.4 (vs. 5.4), Memory (RAM) 5.9 (vs. 5.4), Graphics 4.2 (vs. 3.8), Gaming graphics 3.4 (vs. 3.3), and Primary hard disk 5.7 (vs. 5.4).


After completing this round of data and benchmark collection and comparing, here are my thoughts:
  1. Generally speaking, there is a large performance increase of about 40% ± 5% (depending on the benchmark). It's not surprising considering the 800 MHz increase and doubling of L2 cache of the CPU.
  2. I was a little surprised to see that in both PassMark and Sandra, the test results for the RAM showed a lower latency despite this RAM being 6-6-6-18 rather than 5-5-5-15. I suppose the increased bandwidth of DDR2-800 over DDR2-667 made up for that difference?
  3. I'm not sure why there was a noticeable increase in the graphics scores in PassMark despite the identical chipset used, because I have no experience with integrated graphics.
  4. However, the thing that surprised me were the results for the hard drives. There is a consistent 10%+ advantage of the Seagate drive over the Western Digital drive, and in some cases much more. The reason I'm surprised? Both drive are identical in all the ways you'd think would impact performance. They are both 80 GB capacity drives spinning at 7200 RPM with 8 MB of cache, and both were defragmented prior to testing. My only takeaway from this is that the Barracuda drives were simply better than the Caviar drives of that generation.


Now that I'm finally done with all the testing, it's time to gather the parts for machine #1 and get to work as I need to complete everything by Sunday morning. The next update should happen in the next couple of days.
Warrior1986 is offline  
Sponsored Links
Advertisement
 
post #2 of 29 (permalink) Old 04-26-2014, 08:14 PM - Thread Starter
New to Overclock.net
 
Join Date: Feb 2007
Location: Chicago
Posts: 973
Rep: 26 (Unique: 24)
UPDATE 5/6/14
Ok, now that all the testing is done let's get to putting in some new parts. Here's what machine #1 is undergoing.
  1. CPU - Intel Core 2 Duo E6850 --> Intel Core 2 Duo E8600
  2. GPU - Integrated Intel GMA 3100 --> Galaxy NVIDIA GT 610 GC 1GB GDDR3
  3. Storage - Seagate Barracuda 7200.10 ST380815AS 80GB HDD --> PNY XLR8 240GB SSD


As you can see, a much more modest upgrade intended for a completely different purpose compared to machine #2. Although I did make a point of getting the best dual-core Intel made for socket 775 (the E8700 doesn't count because it was never officially released). I'm not touching the RAM, and the only reason I'm even putting in a discrete GPU is just so mom doesn't have any issues with playback of HD video and to ensure a smooth running interface in any program she's using.

Ready to disassemble.


Optical and floppy drives removed, and HDD caddy moved out of the way. I should note, I did NOT do any cleaning prior to taking these pictures. I was shocked when I opened it up to find it essentially spotless on the inside. I bought some very well-kept systems.


Entire heatsink assembly removed.


No dust between the fins.


Not going to lie, I am impressed with this heatsink. I was quite surprised to find a copper base.


Impressed again to see 2 copper heatpipes.


A Sanyo Denki 80x80x38mm fan serves as both intake and CPU heatsink fan.


Got to have the right tools.


The best socket 775 dual-core CPU is clean and ready to go.


Pin side of the E6850 on the left and the E8600 on the right. Pretty cool to see the results of shrinking down from a 65nm process to a 45nm process.


Base of the heatsink cleaned up.


Tinted the approximate area of contact with the CPU's IHS.


An unintentionally artsy shot of the final install this E8600 will see.


A Core 2 Duo calls for a vertical strip of the TIM.


Reassembled and ready to boot.


Success!


Always a good sign.



The next step will be to run through the same battery of tests. I don't expect there to be as large of a jump in the numbers from a E6850 to a E8600 as there was between the E4500 and E6850, but with the extra 2 MB of L2 cache and addition of SSE4.1 instructions it's hard to say. I'll be lame and guess that there will be a 8-10% increase in the overall system benchmarks solely because of the 10% faster clock speed, and maybe an extra 2% bump to 12% for any CPU-specific benchmarks due to the larger cache and newer SSE instructions. The next update will be of those results. Comparing the results between the E4500, E6850, and E8600 should prove interesting!

UPDATE 5/9/14
All done with testing the E8600.

CPU-Z report - Before & After Results (Click to show)

Wattage use.
Results (Click to show)
Powered off - The same 1 watt as before.

Sitting idle at the desktop - The same 51 watts as before.

Highest load possible using IntelBurnTest. You can see the benefits of going from 65nm to 45nm: it's running 333 MHz faster with 2 MB more L2 cache, but using 6 less watts. It'll probably run a few degrees cooler as well.


Summary of power usage: 1 watt while powered off, 51 watts (E6850 - 51 watts, E4500 - 48 watts) while idle at desktop, 123 watts (E6850 - 129 watts, E4500 - 100 watts) with full CPU load via IntelBurnTest.

Speed of the system's boot up and shutdown.
Results (Click to show)
Cold boot


Restart


Shutdown


Summary of times: 64.3 seconds (E6850 - 60.7 seconds, E4500 - 61.1 seconds) for cold boot, 63.0 seconds (E6850 - 62.3 seconds, E4500 - 65.4 seconds) for restart, 9.0 seconds (E6850 - 8.9 seconds, E4500 - 8.5 seconds) for shutdown.

Benchmark numbers.

Cinebench Results (Click to show)

IntelBurnTest. Once again, you see the power of shrinking die size. This is of course with fresh paste so the max temp will drop even further with time. I don't understand why there is a 4 degree difference between the cores however, as both the E6850 and E4500 had the same max temp for both cores. In addition, this time Core Temp and RealTemp are showing the same temps, whereas with both the E6850 and E4500 RealTemp would show a 5 degree higher temp over Core Temp.
Results (Click to show)
1462MB


3072MB

wPrime Results (Click to show)

CrystalDiskMark Results (Click to show)

CrystalDiskMark 3.0.3 x64 (C) 2007-2013 hiyohiyo
Crystal Dew World : http://crystalmark.info/
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

Sequential Read : 68.981 MB/s
Sequential Write : 67.878 MB/s
Random Read 512KB : 26.677 MB/s
Random Write 512KB : 24.541 MB/s
Random Read 4KB (QD=1) : 0.341 MB/s [ 83.2 IOPS]
Random Write 4KB (QD=1) : 0.816 MB/s [ 199.3 IOPS]
Random Read 4KB (QD=32) : 0.692 MB/s [ 168.8 IOPS]
Random Write 4KB (QD=32) : 0.818 MB/s [ 199.8 IOPS]

Test : 1000 MB [C: 35.3% (26.3/74.4 GB)] (x3)
Date : 2014/05/06 22:56:19
OS : Windows 7 Professional SP1 [6.1 Build 7601] (x64)

PCMark 07 Results (Click to show)

PCMark 08 Results (Click to show)
SiSoftware Sandra. Red is the E8600, blue is the old E6850, and green is machine #2 with the E4500. Results (Click to show)

System Score
Processor Arithmetic
Aggregated Score : 22.82GOPS
Processor Multi-Media
Aggregated Score : 43.21MPix/s
Cryptography
Aggregated Score : 0.438GB/s
Processor Financial Analysis
Aggregated Score : 2.43kOPT/s
.NET Arithmetic
Aggregated Score : 10.68GOPS
.NET Multi-Media
Aggregated Score : 6.04MPix/s
Memory Bandwidth
Aggregated Score : 6.101GB/s
Cache & Memory Latency
Aggregated Score : 47.3ns
File System Bandwidth
Aggregated Score : 51.546MB/s
File System I/O
Aggregated Score : 704.5IOPS
Overall Score
Aggregated Score : 1.32kPT


PassMark PerformanceTest. Green corresponds to the E8600, red to the old E6850, and yellow is machine #2 with the E4500. Results (Click to show)
PassMark Rating


Summary. Link to PassMark's site for my baseline - http://www.passmark.com/baselines/V8/display.php?id=23557161892


CPU Mark


2D Graphics Mark


3D Graphics Mark


Memory Mark


Disk Mark


Windows Experience Index Results (Click to show)

Summary of all benchmarks Results (Click to show)
*Cinebench - 182cb
  • ~22% increase over E6850 score of 149cb
  • ~72% increase over machine #2 E4500 score of 105
*IntelBurnTest (Best of 3 runs) - 1462MB - 68.836 seconds and 22.7027 GFLOPS. 3072MB - 218.776 seconds and 22.6228 GFLOPS
  • ~14% faster/increase over E6850 score of 78.313 seconds and 19.9555 GFLOPS on 1462MB test & ~12% faster/increase over E6850 score of 244.884 seconds and 20.2109 GFLOPS on 3072MB test
  • ~107% faster/increase over machine #2 E4500 score of 142.442 seconds and 10.9713 GFLOPS on 1462MB test
*wPrime 32M test (Best of 3 runs) - 25.989 seconds
  • ~18% faster over E6850 time of 30.623 seconds
  • ~61% faster over machine #2 E4500 time of 42.001 seconds
*CrystalDiskMark with Default Random test date (Best of 3 runs) - Sequential Read 68.981 MB/s, Sequential Write 67.878 MB/s, Random 4KB read @ QD1 of 0.341 MB/s [83.2 IOPS], Random 4KB write @ QD1 of 0.816 MB/s [199.3 IOPS], Random 4KB read @ QD32 of 0.692 MB/s [168.8 IOPS], Random 4KB write @ QD32 of 0.818 MB/s [199.8 IOPS]
  • ~9% faster on sequential (Read 63.289 MB/s, Write 62.419 MB/s), ~10-26% SLOWER on random 4KB QD1 (Read 0.464 MB/s [113.3 IOPS], Write 0.906 MB/s [221.1 IOPS]), and ~14-17% SLOWER on random 4KB QD32 (Read 0.834 MB/s [203.6 IOPS], Write 0.946 MB/s [231.0 IOPS]) over E6850 scores
  • ~24% faster on sequential (Read 55.515 MB/s, Write 53.892 MB/s), ~10% faster on random 4KB QD1 Read 0.313 MB/s [76.4 IOPS], ~5% SLOWER on random 4KB QD1 Write 0.858 MB/s [209.6 IOPS], ~125% faster on random 4KB QD32 Read 0.308 MB/s [75.1 IOPS], ~5% SLOWER on random 4KB QD32 Write 0.865 MB/s [211.3 IOPS] over machine #2 E4500 scores
*PCMark 07 (Best of 3 runs) - 1427
  • ~4.5% increase over E6850 score of 1366
  • ~29.5% increase over machine #2 E4500 score of 1102
*PCMark 08 (Best of 3 runs) - No score due to graphics test failing on both Home 3.0 and Work 2.0 tests
*SiSoftware Sandra Overall Aggregated Score - 1.32kPT
  • ~10% increase over E6850 score of 1.20kPT
  • ~54% increase over machine #2 E4500 score of 0.86kPT
*PassMark PerformanceTest - PassMark rating of 214.5 with subscores of CPU Mark 2764, 2D Graphics Mark 509, 3D Graphics Mark 20.0, Memory Mark 923, and Disk Mark 440.1
  • ~12% increase on PassMark rating (191.6), ~20% increase on CPU Mark (2298), ~16% increase on 2D Graphics Mark (438.8), ~13% increase on 3D Graphics Mark (17.8), ~4% increase on Memory Mark, and ~4% increase on Disk Mark over E6850 scores
  • ~50% increase on PassMark rating (142.9), ~75% increase on CPU Mark (1583), ~73% increase on 2D Graphics Mark (294.5), ~51% increase on 3D Graphics Mark (13.3), ~46% increase on Memory Mark, and 17% increase on Disk Mark over machine #2 E4500 scores
*Windows Experience Index - Base Score of 3.4 with subscores of Processor 6.7, Memory (RAM) 5.9, Graphics 4.2, Gaming graphics 3.4, and Primary hard disk 5.7
  • Base Score of 3.4 with subscores of Processor 6.4, Memory (RAM) 5.9, Graphics 4.2, Gaming graphics 3.4, and Primary hard disk 5.7 on E6850.
  • Base Score of 3.3 with subscores of Processor 5.4, Memory (RAM) 5.4, Graphics 3.8, Gaming graphics 3.3, and Primary hard disk 5.4 on machine #2 E4500.

Thoughts so far:
  1. I was fairly close to my guess of an overall 10% increase in system performance, but the gap for CPU-specific performance was definitely higher than expected when compared to the E6850. And no surprise that the E4500 doesn't even come close to the E8600. So anyone who had a E4000 series CPU could spend $40 on eBay for a E8600 and dramatically increase the overall speed of the system.
  2. It appears to have been confirmed by both PassMark and CrystalDiskMark, and especially in Sandra, that drive performance DROPPED. While sequential speeds may have gone up slightly, all 3 tests showed random data access IOPS getting hampered with Sandra showing a massive 50% decrease in IOPS. How is it possible that putting in a faster CPU would negatively impact IOPS?

It's now crunchtime with just under a day left for me to finish up this machine. The plan for tomorrow will be to install the GPU and only run PCMark 07, 08, Sandra, and PassMark, then post those results comparing only to the scores I just got after putting in the E8600. After that, the SSD will go in and I'll get the system up and running and fully optimized on the fresh install of Win 7 Ultimate 64-bit, at which point I'll run the full battery of tests and compare to scores prior to installing the SSD.

UPDATE 5/15/14
Well as I expected, time was a huge issue. I had to stay up VERY late on Saturday to get the machine done, as I forgot that doing all the initial updating and setup takes a while, not to mention all the optimizations and tweaking I needed to do. Anyways, on to some pictures.

Here's what the regular/normal Galaxy Nvidia Geforce GT 610 GC looks like.


And this one is mine. Wahhh?!?


Side view.


Proof it is in fact a GT 610.


So what's going on here? Well when I picked up the card and installed it, I noticed it had an incredibly noisy fan. I thought maybe it was a defective fan so I swapped it for another one and it had the exact same annoying noise coming from it. Also, the idle temps were very high sitting right at about 41-42 degrees C. I was about to say forget it and just return the card, but I looked at the GT 620 and noticed the card's had a similar layout and the mounting holes matched up so I had an idea. The result of that idea is what you see above, probably one of a handful of GT 610 in existence with the cooler from the GT 620. The end result? MASSIVE reduction in noise. You literally cannot hear it above the sound of the front intake fan unless you put your ear right up to the card, and even then it's just a barely audible whirring sound typical of small fans. An unintended side benefit was the huge drop in idle temp of about 10-11 degrees C (verified through GPU-Z) to 31-32 degrees C.

First off we take a look at the cooler I removed.


As you can see, the fan is pathetically small with tiny and thin fin blades.


A very cheap heatsink. I'm honestly not convinced it was even 100% metal, given the awful idle temps.


NOW we are talking.


Much bigger fan with better designed fin blades.


THIS is a proper heatsink.


Here's the problem. Different plugs for some stupid reason despite the fact they both only use 2 wires.



So what did I do? I took off the plug bracket and just left the bare prongs. Works perfectly.


I did have to modify the angle of one of the fins.


This is why: a lone capacitor that is exactly where the fin would have been.



Even with the modification, I'm not comfortable with the proximity of the capacitor to the card. So a little strip of electrical tape for insurance.


Putting the SSD in. Nice quality metal shell and already has the latest firmware on it (for some reason, the latest one on PNY's website is 5.0.7 confused.gif)



So this machine is completed, and is up and running and being used daily by my mom. I will be doing a final round of testing on it, and will compare it to her old machine. Even without the results yet though, I can already tell it's so much faster. Boot time has been cut by at LEAST half, and Firefox is running without hiccups with dozens of tabs open. She is incredibly happy with it and said to me "How is this computer so fast now? What wizardry did you do?" HAH!

Once I get all the testing done, I'll post up the results. Meanwhile I finally was able to do a little work on machine #2.

One more shot before they go in.


Pulled the old sticks out and all clear for 8GB.


Snug as a bug (literally, took some pressure to get the clips to lock down).


Before.


After. I'm not sure why it changed from rank 1 to 2 nor what that means.


All 8GB detected.


Top row is before and bottom row is after. I was glad to see the tighter timings of new RAM were JEDEC-compliant, as the BIOS has no options for manual settings.


That wraps up this update. Next one will be the final results and comparisons for machine #1, testing machine #2 with the new RAM, installing the new CPU, or possibly some combination.
Warrior1986 is offline  
post #3 of 29 (permalink) Old 04-26-2014, 08:14 PM - Thread Starter
Warrior1986 is offline  
Sponsored Links
Advertisement
 
post #4 of 29 (permalink) Old 04-26-2014, 08:14 PM - Thread Starter
Warrior1986 is offline  
post #5 of 29 (permalink) Old 04-26-2014, 08:15 PM - Thread Starter
Warrior1986 is offline  
post #6 of 29 (permalink) Old 04-26-2014, 08:15 PM - Thread Starter
Warrior1986 is offline  
post #7 of 29 (permalink) Old 04-26-2014, 08:15 PM - Thread Starter
Warrior1986 is offline  
post #8 of 29 (permalink) Old 05-03-2014, 05:10 PM - Thread Starter
New to Overclock.net
 
Join Date: Feb 2007
Location: Chicago
Posts: 973
Rep: 26 (Unique: 24)
Warrior1986 is offline  
post #9 of 29 (permalink) Old 05-05-2014, 02:50 PM - Thread Starter
Warrior1986 is offline  
post #10 of 29 (permalink) Old 05-05-2014, 02:53 PM
New to Overclock.net
 
Dyson Poindexter's Avatar
 
Join Date: Jun 2009
Posts: 3,638
Rep: 209 (Unique: 161)
This is way cool! I love the SFF Dells, they're actually quite clever with how they lay them out.

Why do we mount round fans on square radiators?


Dyson Poindexter is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off