Highest MIPS - screenshots of 7-zip built-in benchmark - Page 2 - Overclock.net - An Overclocking Community

Forum Jump: 

Highest MIPS - screenshots of 7-zip built-in benchmark

Reply
 
Thread Tools
post #11 of 69 (permalink) Old 01-29-2018, 06:32 PM - Thread Starter
Integer Benchmarker
 
Sanmayce's Avatar
 
Join Date: Mar 2012
Location: Sofia
Posts: 392
Rep: 22 (Unique: 19)
Thanks, it is much better.
Utilization now is 3160%, as it should.

Code:
---------------------------------------------------------------------------------------------------------------------
| OCN username | CPU                | Threads | CPU frequency OC | RAM frequency OC | RAM (CL) | Decompression MIPS |
---------------------------------------------------------------------------------------------------------------------
| Jpmboy       | Core i9 7980XE     | 36      | 4.5GHz           | 4000MHz          | 16       | 114481             |
| Tamalero     | Threadripper 1950X | 32      | 3.9GHz           | 3066MHz          | 16       | 99717              |
| gupsterg     | Threadripper 1950X | 32      | ?GHz             | 3200MHz          | 14       | 94081              |
| DStealth     | Core i7 8700K      | 12      | 5552MHz          | 4000MHz          | 16       | 49426              |
| Diablosbud   | Ryzen 5 1600       | 12      | 3.875 GHz        | 3200MHz          | 14       | 36784              |
| Sanmayce     | Core i5 7200U      | 4       | 2712MHz          | 2133MHz          | 15       | 9056               |
---------------------------------------------------------------------------------------------------------------------

Get down get down get down get it on show love and give it up
What are you waiting on?
Sanmayce is offline  
Sponsored Links
Advertisement
 
post #12 of 69 (permalink) Old 01-29-2018, 11:51 PM
Meddling user
 
gupsterg's Avatar
 
Join Date: Jan 2015
Location: Lurking over a keyboard
Posts: 6,952
Rep: 741 (Unique: 345)
Sanmayce yes you have the info now on TR . I was at 3.95GHz, AIDA64 screenie shows this .

This is your thread/results gathering , so how you wish to do is your choice . I believe you are using the Current reading from Decompressing of Rating in MIPS. I can stop benchmark at differing time in a run to show an improved result for same settings. I think the Resulting fields are more representative of true performance for setup. See below screenies on same settings, without reposting, etc.
Attached Thumbnails
Click image for larger version

Name:	7-Zip 97665.jpg
Views:	27
Size:	234.6 KB
ID:	64249  

Click image for larger version

Name:	7-Zip 99041.jpg
Views:	10
Size:	128.5 KB
ID:	64257  

Click image for larger version

Name:	7-Zip 100067.jpg
Views:	17
Size:	237.4 KB
ID:	64265  


Last edited by gupsterg; 01-29-2018 at 11:56 PM.
gupsterg is offline  
post #13 of 69 (permalink) Old 01-30-2018, 06:58 AM - Thread Starter
Integer Benchmarker
 
Sanmayce's Avatar
 
Join Date: Mar 2012
Location: Sofia
Posts: 392
Rep: 22 (Unique: 19)
>I was at 3.95GHz, AIDA64 screenie shows this .

My bad, sometimes (too often recently) tunnel vision is my mode, stupid.

>This is your thread/results gathering , so how you wish to do is your choice .

I don't feel that way, always open for new suggestions/layouts/approaches. Did choose Decompression MIPS as a mastermetric because of nearly full utilization of all the threads, not so with Compression, yet, the latter is even more important/indicative for power of one CPU-RAM subsystem. So, both are to be used in the future.

>I believe you are using the Current reading from Decompressing of Rating in MIPS.

Yes, but if 'Current' is lower than 'Result' then the bigger enters the roster.

>I can stop benchmark at differing time in a run to show an improved result for same settings. I think the Resulting fields are more representative of true performance for setup.

I would write to Igor to consider printing more stable stats, as:
- Initial (non-warmt) i.e. 1st pass stats;
- Average (all the runs summed and divided by their number);
- Top stats - the highest value of all runs.

The third one, to me, is absolutely a must-have.
However, the "legacy" is a heavy burden and my guess is no one will be gladdened to rehash all the gathered data, except me.

Code:
---------------------------------------------------------------------------------------------------------------------
| OCN username | CPU                | Threads | CPU frequency OC | RAM frequency OC | RAM (CL) | Decompression MIPS |
---------------------------------------------------------------------------------------------------------------------
| Jpmboy       | Core i9 7980XE     | 36      | 4.5GHz           | 4000MHz          | 16       | 114481             |
| gupsterg     | Threadripper 1950X | 32      | 3.95GHz          | 3200MHz          | 14       | 100067             |
| Tamalero     | Threadripper 1950X | 32      | 3.9GHz           | 3066MHz          | 16       | 99717              |
| DStealth     | Core i7 8700K      | 12      | 5552MHz          | 4000MHz          | 16       | 49426              |
| Diablosbud   | Ryzen 5 1600       | 12      | 3.875 GHz        | 3200MHz          | 14       | 36784              |
| Sanmayce     | Core i5 7200U      | 4       | 2712MHz          | 2133MHz          | 15       | 9056               |
---------------------------------------------------------------------------------------------------------------------
Wanna address few more moments:

I always compare Decompression stats instead of Compression, the utilization of the former is nearly the number of threads x 100%, to draw conclusions it should be that way.
As Igor stated in his homepage, he tried to run more threads in 7-zip compression (than the actually available) in order to saturate the load.
So, compression stats are not that indicative! Yet, very important.

The screenshot with the 100067 MIPS is achieved at 3078% utilization (with 120/1772 Processes/Threads), room for betterment, on top of that in order to reach for highest output the benchmark should be run as:
- Administrator;
- In Realtime priority.
My point, we as users, still don't have a stable one-stop-shop reporter of [de]compression performance.


Attached Thumbnails
Click image for larger version

Name:	7200_b.png
Views:	380
Size:	657.8 KB
ID:	64417  

Click image for larger version

Name:	7200_a.png
Views:	369
Size:	665.8 KB
ID:	64425  


Get down get down get down get it on show love and give it up
What are you waiting on?
Sanmayce is offline  
Sponsored Links
Advertisement
 
post #14 of 69 (permalink) Old 02-03-2018, 06:45 PM
New to Overclock.net
 
Tamalero's Avatar
 
Join Date: Dec 2016
Posts: 322
Rep: 5 (Unique: 5)
For testing purposes just tried the same bench a few times at 4Ghz.
I guess there isnt that much of a difference. I believe the RAM affects the scores a lot too.

https://www.overclock.net/forum/attac...=71825&thumb=1
Attached Thumbnails
Click image for larger version

Name:	jQPrcdbc.jpg
Views:	22
Size:	144.7 KB
ID:	71825  

Tamalero is offline  
post #15 of 69 (permalink) Old 02-03-2018, 08:00 PM - Thread Starter
Integer Benchmarker
 
Sanmayce's Avatar
 
Join Date: Mar 2012
Location: Sofia
Posts: 392
Rep: 22 (Unique: 19)
Thanks, you are right, RAM settings matter a lot, yet, the huge discrepancy 113234KB/s (Core i9 7980XE) vs 60,145KB/s (Threadripper 1950X) still puzzles me.

Code:
---------------------------------------------------------------------------------------------------------------------
| OCN username | CPU                | Threads | CPU frequency OC | RAM frequency OC | RAM (CL) | Decompression MIPS |
---------------------------------------------------------------------------------------------------------------------
| Jpmboy       | Core i9 7980XE     | 36      | 4.5GHz           | 4000MHz          | 16       | 114481             |
| Tamalero     | Threadripper 1950X | 32      | 4.0GHz           | 3066MHz          | 16       | 100579             |
| gupsterg     | Threadripper 1950X | 32      | 3.95GHz          | 3200MHz          | 14       | 100067             |
| DStealth     | Core i7 8700K      | 12      | 5552MHz          | 4000MHz          | 16       | 49426              |
| Diablosbud   | Ryzen 5 1600       | 12      | 3.875 GHz        | 3200MHz          | 14       | 36784              |
| Sanmayce     | Core i5 7200U      | 4       | 2712MHz          | 2133MHz          | 15       | 9056               |
---------------------------------------------------------------------------------------------------------------------
Also, the delta is too big: 96829 MIPS (at first run) versus 100579 MIPS (at 8th run). Those MIPS (million instructions per second) should be more stable. In essence, IPC is what mostly matters. Don't know why, but I cannot find any benchmark reporting IPC during heavy IMCs (Internal Memory ControllerS) loads, meaning bombardment of RAM with random 256bit reads (mostly uncached), hope 'Freaky Dreamer' package proves useful in tweaking 32-threaded CPUs with RAM various settings:
https://www.overclock.net/forum/21-be...l#post26663593

Regardless of different frequencies/steppings at which cores operate we need a stable mastermetric, namely, Instructions-Per-Clock.

Get down get down get down get it on show love and give it up
What are you waiting on?
Sanmayce is offline  
post #16 of 69 (permalink) Old 02-05-2018, 11:59 AM
New to Overclock.net
 
Join Date: Apr 2012
Posts: 884
Rep: 42 (Unique: 38)
4.5 on chip, ram @ 4000, C16
Attached Thumbnails
Click image for larger version

Name:	mips.jpg
Views:	67
Size:	349.5 KB
ID:	73065  

DooRules is offline  
post #17 of 69 (permalink) Old 02-05-2018, 02:21 PM - Thread Starter
Integer Benchmarker
 
Sanmayce's Avatar
 
Join Date: Mar 2012
Location: Sofia
Posts: 392
Rep: 22 (Unique: 19)
Nice, thank you DooRules, your machine rules, indeed!

The cute RED superpig (a popular Chinese mascot) has its ear pulled, reminds me of AMD's supercpu - lovely but thrashed.

Code:
---------------------------------------------------------------------------------------------------------------------
| OCN username | CPU                | Threads | CPU frequency OC | RAM frequency OC | RAM (CL) | Decompression MIPS |
---------------------------------------------------------------------------------------------------------------------
| DooRules     | Core i9 7980XE     | 36      | 4.5GHz           | 4000MHz          | 16       | 119726             |
| Jpmboy       | Core i9 7980XE     | 36      | 4.5GHz           | 4000MHz          | 16       | 114481             |
| Tamalero     | Threadripper 1950X | 32      | 4.0GHz           | 3066MHz          | 16       | 100579             |
| gupsterg     | Threadripper 1950X | 32      | 3.95GHz          | 3200MHz          | 14       | 100067             |
| DStealth     | Core i7 8700K      | 12      | 5552MHz          | 4000MHz          | 16       | 49426              |
| Diablosbud   | Ryzen 5 1600       | 12      | 3.875 GHz        | 3200MHz          | 14       | 36784              |
| Sanmayce     | Core i5 7200U      | 4       | 2712MHz          | 2133MHz          | 15       | 9056               |
---------------------------------------------------------------------------------------------------------------------
However, comparing your Compression MIPS to Jpmboy's, it is second:
113234 KB/s (Core i9 7980XE)
-
109318 KB/s (Core i9 7980XE)
4MB/s
The access time to uncached RAM is roughly 2x smaller when compared to Threadripper, hence (AFAIU) the nearly 2x better Compression rates:
"Compression speed strongly depends from memory (RAM) latency, Data Cache size/speed and TLB. Out-of-Order execution feature of CPU is also important for that test."
Source: http://www.7-cpu.com/

My dissatisfaction from 7-zip fluctuations stands, your two machines cannot be "rated" in definite way. The whole point of [de]compression stats is to "catch" all those tweaks/settings of CPU-RAM subsystem - to see their affects, 7-zip is good but not enough.

Get down get down get down get it on show love and give it up
What are you waiting on?

Last edited by Sanmayce; 02-05-2018 at 03:30 PM.
Sanmayce is offline  
post #18 of 69 (permalink) Old 02-06-2018, 03:56 PM
New to Overclock.net
 
Join Date: Apr 2012
Posts: 884
Rep: 42 (Unique: 38)
was playing around with something else @ 4,7, gave this a shot
Attached Thumbnails
Click image for larger version

Name:	4.7.PNG
Views:	51
Size:	180.7 KB
ID:	74601  

DooRules is offline  
post #19 of 69 (permalink) Old 02-06-2018, 04:10 PM - Thread Starter
Integer Benchmarker
 
Sanmayce's Avatar
 
Join Date: Mar 2012
Location: Sofia
Posts: 392
Rep: 22 (Unique: 19)
Many thanks, very interesting.

The first thing that caught my attention was the jump in AIDA's brutally fast L1 cache bandwidth getting even more brutal:
from 4972 GB/s (@4500MHz) to 5188 GB/s (@4700MHz)

The second thing was the nearly maximum utilization during your Decompression - 3588% - only 12% below the 36x100% limit! Such results I call MUTSI, that is, really indicative!

Code:
---------------------------------------------------------------------------------------------------------------------
| OCN username | CPU                | Threads | CPU frequency OC | RAM frequency OC | RAM (CL) | Decompression MIPS |
---------------------------------------------------------------------------------------------------------------------
| DooRules     | Core i9 7980XE     | 36      | 4.7GHz           | 4000MHz          | 16       | 125571             |
| DooRules     | Core i9 7980XE     | 36      | 4.5GHz           | 4000MHz          | 16       | 119726             |
| Jpmboy       | Core i9 7980XE     | 36      | 4.5GHz           | 4000MHz          | 16       | 114481             |
| Tamalero     | Threadripper 1950X | 32      | 4.0GHz           | 3066MHz          | 16       | 100579             |
| gupsterg     | Threadripper 1950X | 32      | 3.95GHz          | 3200MHz          | 14       | 100067             |
| DStealth     | Core i7 8700K      | 12      | 5552MHz          | 4000MHz          | 16       | 49426              |
| Diablosbud   | Ryzen 5 1600       | 12      | 3.875 GHz        | 3200MHz          | 14       | 36784              |
| Sanmayce     | Core i5 7200U      | 4       | 2712MHz          | 2133MHz          | 15       | 9056               |
---------------------------------------------------------------------------------------------------------------------
Very glad that you shared these very informative shots.

Get down get down get down get it on show love and give it up
What are you waiting on?
Sanmayce is offline  
post #20 of 69 (permalink) Old 05-08-2018, 11:04 AM
New to Overclock.net
 
RealSteelH6's Avatar
 
Join Date: Oct 2015
Posts: 183
Rep: 7 (Unique: 7)
Intel Core i5 [email protected]
G.Skill Trident Z RGB Kit 16GB DDR4-3200 CL14
MSI Z370M Mortar
Attached Thumbnails
Click image for larger version

Name:	7z.PNG
Views:	73
Size:	74.1 KB
ID:	180225  


CPU: Intel Core i5-8600K @ 4600MHz
RAM: G.Skill Trident Z RGB 16GB @ 3200MHz
GPU: Nvidia GeForce GTX 1070 Ti FE
SSD: Samsung 850 Evo @ 500 Gb
MOU: SteelSeries Rival 110
PAD: ASUS ROG Sheath
KEY: Logitech G213
MON: Dell S2417DG
RealSteelH6 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