Overclock.net › Forums › Components › Memory › Am i Correct?
New Posts  All Forums:Forum Nav:

Am i Correct? - Page 3

post #21 of 34
okay, i decided to shut this topic down once and for all, as it keeps coming back up again every few days. unfortunately, i can't get a 1:2 ratio with my memory, so it will take someone with some uber sticks of RAM to really illustrate this point.

i lowered my FSB all the way down to 266 so i could run some benchmarks with my memory at DDR 533 (1:1 ratio) and DDR 800 (2:3 ratio). like i said, 1:2 ratio would be better but i don't think my memory can handle DDR 1066. memory timings were at 4-4-3-5 2T for both sets of tests.

seeing as how i'm lazy and unimaginative, i only ran three sets of benchmarks: PCMark05, the Multicore Efficiency test on SiSandra, and the Processor Arithmetic test even though i knew the last one wouldn't show anything.

PCMark05
it should be noted that PCMark tests all around system performance, so the total score difference can't be compared by percentage points. for example, parts of the test stress the video card, and those scores were obviously the same regardless of my memory divider.

1:1 Ratio Score = 5,868
2:3 Ratio Score = 6,072

not a huge improvement with the memory divider, but the part of the test that focuses on memory latency showed a 31% improvement with the divider. i photoshopped my score details together with the 2:3 ratio score on the right:



and remember i couldn't even achieve a 1:2 ratio like i wanted to. everything was the same on the second run except for my memory frequency. 31% more accesses per second.

the second test i ran, as an afterthought really, was the Multicore Efficiency test on SiSandra.

1:1 Ratio Score = 5814 MB/s
Latency = 82ms

2:3 Ratio Score = 6072MB/s
Latency = 82ms

just a 4% increase with the memory divider. i'd want to run that test a few more times to see if those scores are consistent. either way, it wasn't a huge difference.

i also ran the Processor Arithmetic test afterward and, not surprisingly, my MIPS and MFLOPS were the same with both memory settings. just like SuperPI, that test doesn't appear to be affected by memory speed much.


so... i guess the jury is still out until someone with better memory can continue these tests. that 31% increase in the memory score on PCMark05 really grabbed my attention though. those of you saying 1:1 is as good as it gets might want to do some tests yourself.
The C'roe
(13 items)
 
  
CPUMotherboardGraphicsRAM
E6320 @ 2.8 Asus P5W DH Deluxe (2) Radeon X1900XTX CF 2GB G.Skill HK 4-4-3-5 
Hard DriveOptical DriveOSMonitor
640 - 2x320 Samsung RAID-0 Dual DVDs XP MCE Acer 22" + Dell 19" LCDs - 2960x1024 rez :D 
KeyboardPowerCaseMouse
Logitech MX Zalman 600W RT Modified Antec 900 Logitech MX 
Mouse Pad
Giant 
  hide details  
Reply
The C'roe
(13 items)
 
  
CPUMotherboardGraphicsRAM
E6320 @ 2.8 Asus P5W DH Deluxe (2) Radeon X1900XTX CF 2GB G.Skill HK 4-4-3-5 
Hard DriveOptical DriveOSMonitor
640 - 2x320 Samsung RAID-0 Dual DVDs XP MCE Acer 22" + Dell 19" LCDs - 2960x1024 rez :D 
KeyboardPowerCaseMouse
Logitech MX Zalman 600W RT Modified Antec 900 Logitech MX 
Mouse Pad
Giant 
  hide details  
Reply
post #22 of 34
That because the timmings were too close to each other then of course the higher mhz memory was better because higher mhz with the same timmings will result in less nanoseconds between each data transfer.

Past a certain point (usualy the cutoff between 1T and 2T) the gain you get with higher mhz are less than what tighter timmings give you.

Until DDR3 with good timmings or very high frequency come out then that will stay the case since DDR that reach higher than 1200Mhz is hard to come by anyway. If 1300-1400Mhz DDR2 was available with respectable timmings (5-5-5) then yeah higher mhz would take back the first place because the timming and number of hz calculated together = a number in nanosecond lower than say 900mhz with tighter timmings.

I will try this test because my ram can do 1066. Once I am done with my other things.
Cerebro
(13 items)
 
  
CPUMotherboardGraphicsRAM
C2D E6420 3.4Ghz (8x425FSB 1:1) Evga 680i SLI NF68-T1 - P31 Asus Geforce 8800GTX OCZ Titanium Alpha VX2 + Crucial Ballistix PC8500 
Hard DriveOptical DriveOSMonitor
7 Hard Drives - All 7200RPM External LG DVD-Writer 16x via Firewire Windows XP SP2 Compaq V1000 21" CRT 
KeyboardPowerCaseMouse
Logitech G15 System - Enermax 565W / Hard Drives - Enermax 350W Thermaltake Mozart TX Logitech MX1000 
  hide details  
Reply
Cerebro
(13 items)
 
  
CPUMotherboardGraphicsRAM
C2D E6420 3.4Ghz (8x425FSB 1:1) Evga 680i SLI NF68-T1 - P31 Asus Geforce 8800GTX OCZ Titanium Alpha VX2 + Crucial Ballistix PC8500 
Hard DriveOptical DriveOSMonitor
7 Hard Drives - All 7200RPM External LG DVD-Writer 16x via Firewire Windows XP SP2 Compaq V1000 21" CRT 
KeyboardPowerCaseMouse
Logitech G15 System - Enermax 565W / Hard Drives - Enermax 350W Thermaltake Mozart TX Logitech MX1000 
  hide details  
Reply
post #23 of 34
Hmm, but isn't the point of running slower RAM to have tighter timings? So shouldn't you also run a test with a 1:1 ratio having tighter timings and a 2:3 with more loose timings?


Edit: Do'h Cyant beat me to it.
X-Dreamer
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core 2 Duo E6600 EVGA 680i EVGA 8800GTX 4x1GB Ballistix DDR2-800 
Hard DriveOSMonitorPower
500GB Samsung Windows Vista Home Premium Samsung 226BW 22 inch LCD Thermaltake toughpower 750W 
Case
Turbo X-Dreamer II 
  hide details  
Reply
X-Dreamer
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core 2 Duo E6600 EVGA 680i EVGA 8800GTX 4x1GB Ballistix DDR2-800 
Hard DriveOSMonitorPower
500GB Samsung Windows Vista Home Premium Samsung 226BW 22 inch LCD Thermaltake toughpower 750W 
Case
Turbo X-Dreamer II 
  hide details  
Reply
post #24 of 34
this thread wasn't about timings at all. many people are saying 1:1 gives you the best possible performance. some of us are arguing that you can see increased performance if you can clock your memory higher than your FSB. the other guys are saying to stop at 1:1 because that's as good as it gets.

and i'm not arguing a point to show off my memory - i haven't been able to get mine over DDR2 850. i wish i could get it higher.
The C'roe
(13 items)
 
  
CPUMotherboardGraphicsRAM
E6320 @ 2.8 Asus P5W DH Deluxe (2) Radeon X1900XTX CF 2GB G.Skill HK 4-4-3-5 
Hard DriveOptical DriveOSMonitor
640 - 2x320 Samsung RAID-0 Dual DVDs XP MCE Acer 22" + Dell 19" LCDs - 2960x1024 rez :D 
KeyboardPowerCaseMouse
Logitech MX Zalman 600W RT Modified Antec 900 Logitech MX 
Mouse Pad
Giant 
  hide details  
Reply
The C'roe
(13 items)
 
  
CPUMotherboardGraphicsRAM
E6320 @ 2.8 Asus P5W DH Deluxe (2) Radeon X1900XTX CF 2GB G.Skill HK 4-4-3-5 
Hard DriveOptical DriveOSMonitor
640 - 2x320 Samsung RAID-0 Dual DVDs XP MCE Acer 22" + Dell 19" LCDs - 2960x1024 rez :D 
KeyboardPowerCaseMouse
Logitech MX Zalman 600W RT Modified Antec 900 Logitech MX 
Mouse Pad
Giant 
  hide details  
Reply
post #25 of 34
Alright I did some benching to pove my point here are the screenshots:

266:266 1T


266:400 2T


266:400 1T


266:533 2T


304:266 1T


304:304 1T


304:408 2T


All the 266Mhz FSB scores are more or less same or within the margin of error anyway. The 1:1 ratio has the best score actually.

Now let's increase the FSB by 15% and lower the multi to 7x and we get the same number of Mhz for the CPU but a 15% faster FSB.

With a 304Mhz FSB and 266Mhz memory the memory read speed is actually the same as 266Mhz FSB memory read (7.2-7.4k range). Once the FSB:Memory speed is 1:1 then the score increase to the 8.2K range. if we raise the memory speed further the score stay in the 8K range (actually slower than 1:1 again tho).

This tell us that the "Memory Read speed" is limited by the FSB speed OR the Memory speed WITCHEVER is slower.

The memory write and copy is not affected by the memory speed in the tests. As long as the memory is at least 1:1
Cerebro
(13 items)
 
  
CPUMotherboardGraphicsRAM
C2D E6420 3.4Ghz (8x425FSB 1:1) Evga 680i SLI NF68-T1 - P31 Asus Geforce 8800GTX OCZ Titanium Alpha VX2 + Crucial Ballistix PC8500 
Hard DriveOptical DriveOSMonitor
7 Hard Drives - All 7200RPM External LG DVD-Writer 16x via Firewire Windows XP SP2 Compaq V1000 21" CRT 
KeyboardPowerCaseMouse
Logitech G15 System - Enermax 565W / Hard Drives - Enermax 350W Thermaltake Mozart TX Logitech MX1000 
  hide details  
Reply
Cerebro
(13 items)
 
  
CPUMotherboardGraphicsRAM
C2D E6420 3.4Ghz (8x425FSB 1:1) Evga 680i SLI NF68-T1 - P31 Asus Geforce 8800GTX OCZ Titanium Alpha VX2 + Crucial Ballistix PC8500 
Hard DriveOptical DriveOSMonitor
7 Hard Drives - All 7200RPM External LG DVD-Writer 16x via Firewire Windows XP SP2 Compaq V1000 21" CRT 
KeyboardPowerCaseMouse
Logitech G15 System - Enermax 565W / Hard Drives - Enermax 350W Thermaltake Mozart TX Logitech MX1000 
  hide details  
Reply
post #26 of 34
Dang... this debate is everywhere... someone is trying to prove it on the CPU threads too.
Smilodon
(13 items)
 
  
CPUMotherboardGraphicsRAM
e4400 @ 2.9Ghz EVGA 650i Ultra ATX MSI 8800GTS 640 2x1GB G.Skill DDR2 800 
Hard DriveOSMonitorKeyboard
Seagate Barracuda 320Gb Vista 32bit 32" LCD (Vizio) Logitech G15 
PowerCaseMouse
Corsair HX520w Raidmax Smilodon Logitech MX518 
  hide details  
Reply
Smilodon
(13 items)
 
  
CPUMotherboardGraphicsRAM
e4400 @ 2.9Ghz EVGA 650i Ultra ATX MSI 8800GTS 640 2x1GB G.Skill DDR2 800 
Hard DriveOSMonitorKeyboard
Seagate Barracuda 320Gb Vista 32bit 32" LCD (Vizio) Logitech G15 
PowerCaseMouse
Corsair HX520w Raidmax Smilodon Logitech MX518 
  hide details  
Reply
post #27 of 34
first of all, Image Shack sucks. i can't read any of your screenies because the page just re-loads when i click the image to enlarge it.

second, K.I.S.S. don't raise your memory timings when you raise the frequency. you could just end up cancelling out the speed you just added to it. also, stop changing your CPU freq. and/or multi. keep everything the same but the memory speed. if you have to run loose timings at the lower speed, so be it. keep everything the same but the memory speed.
The C'roe
(13 items)
 
  
CPUMotherboardGraphicsRAM
E6320 @ 2.8 Asus P5W DH Deluxe (2) Radeon X1900XTX CF 2GB G.Skill HK 4-4-3-5 
Hard DriveOptical DriveOSMonitor
640 - 2x320 Samsung RAID-0 Dual DVDs XP MCE Acer 22" + Dell 19" LCDs - 2960x1024 rez :D 
KeyboardPowerCaseMouse
Logitech MX Zalman 600W RT Modified Antec 900 Logitech MX 
Mouse Pad
Giant 
  hide details  
Reply
The C'roe
(13 items)
 
  
CPUMotherboardGraphicsRAM
E6320 @ 2.8 Asus P5W DH Deluxe (2) Radeon X1900XTX CF 2GB G.Skill HK 4-4-3-5 
Hard DriveOptical DriveOSMonitor
640 - 2x320 Samsung RAID-0 Dual DVDs XP MCE Acer 22" + Dell 19" LCDs - 2960x1024 rez :D 
KeyboardPowerCaseMouse
Logitech MX Zalman 600W RT Modified Antec 900 Logitech MX 
Mouse Pad
Giant 
  hide details  
Reply
post #28 of 34
Then your browser suck because they are resizing properly on my PC and on my Mac. But Image shack is quite slow today thats the only problem I have with them.
Cerebro
(13 items)
 
  
CPUMotherboardGraphicsRAM
C2D E6420 3.4Ghz (8x425FSB 1:1) Evga 680i SLI NF68-T1 - P31 Asus Geforce 8800GTX OCZ Titanium Alpha VX2 + Crucial Ballistix PC8500 
Hard DriveOptical DriveOSMonitor
7 Hard Drives - All 7200RPM External LG DVD-Writer 16x via Firewire Windows XP SP2 Compaq V1000 21" CRT 
KeyboardPowerCaseMouse
Logitech G15 System - Enermax 565W / Hard Drives - Enermax 350W Thermaltake Mozart TX Logitech MX1000 
  hide details  
Reply
Cerebro
(13 items)
 
  
CPUMotherboardGraphicsRAM
C2D E6420 3.4Ghz (8x425FSB 1:1) Evga 680i SLI NF68-T1 - P31 Asus Geforce 8800GTX OCZ Titanium Alpha VX2 + Crucial Ballistix PC8500 
Hard DriveOptical DriveOSMonitor
7 Hard Drives - All 7200RPM External LG DVD-Writer 16x via Firewire Windows XP SP2 Compaq V1000 21" CRT 
KeyboardPowerCaseMouse
Logitech G15 System - Enermax 565W / Hard Drives - Enermax 350W Thermaltake Mozart TX Logitech MX1000 
  hide details  
Reply
post #29 of 34
this isn't about Mac vs. PC, but i know the problem is on my end. i think it has to do with my pop-up blocker. when i click to enlarge the image, the stupid website tries to give me a pop-up ad and i think that is why my page reloads.

anyway, Photobucket is so much better and faster.
The C'roe
(13 items)
 
  
CPUMotherboardGraphicsRAM
E6320 @ 2.8 Asus P5W DH Deluxe (2) Radeon X1900XTX CF 2GB G.Skill HK 4-4-3-5 
Hard DriveOptical DriveOSMonitor
640 - 2x320 Samsung RAID-0 Dual DVDs XP MCE Acer 22" + Dell 19" LCDs - 2960x1024 rez :D 
KeyboardPowerCaseMouse
Logitech MX Zalman 600W RT Modified Antec 900 Logitech MX 
Mouse Pad
Giant 
  hide details  
Reply
The C'roe
(13 items)
 
  
CPUMotherboardGraphicsRAM
E6320 @ 2.8 Asus P5W DH Deluxe (2) Radeon X1900XTX CF 2GB G.Skill HK 4-4-3-5 
Hard DriveOptical DriveOSMonitor
640 - 2x320 Samsung RAID-0 Dual DVDs XP MCE Acer 22" + Dell 19" LCDs - 2960x1024 rez :D 
KeyboardPowerCaseMouse
Logitech MX Zalman 600W RT Modified Antec 900 Logitech MX 
Mouse Pad
Giant 
  hide details  
Reply
post #30 of 34
Does photobucket have a Firefox addon / Browser toolbar to easily upload images like image shack?

Try pressing ALT or CTRL or something while clicking the image
Cerebro
(13 items)
 
  
CPUMotherboardGraphicsRAM
C2D E6420 3.4Ghz (8x425FSB 1:1) Evga 680i SLI NF68-T1 - P31 Asus Geforce 8800GTX OCZ Titanium Alpha VX2 + Crucial Ballistix PC8500 
Hard DriveOptical DriveOSMonitor
7 Hard Drives - All 7200RPM External LG DVD-Writer 16x via Firewire Windows XP SP2 Compaq V1000 21" CRT 
KeyboardPowerCaseMouse
Logitech G15 System - Enermax 565W / Hard Drives - Enermax 350W Thermaltake Mozart TX Logitech MX1000 
  hide details  
Reply
Cerebro
(13 items)
 
  
CPUMotherboardGraphicsRAM
C2D E6420 3.4Ghz (8x425FSB 1:1) Evga 680i SLI NF68-T1 - P31 Asus Geforce 8800GTX OCZ Titanium Alpha VX2 + Crucial Ballistix PC8500 
Hard DriveOptical DriveOSMonitor
7 Hard Drives - All 7200RPM External LG DVD-Writer 16x via Firewire Windows XP SP2 Compaq V1000 21" CRT 
KeyboardPowerCaseMouse
Logitech G15 System - Enermax 565W / Hard Drives - Enermax 350W Thermaltake Mozart TX Logitech MX1000 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Memory
Overclock.net › Forums › Components › Memory › Am i Correct?