Author Topic: Anomolies of different PC timings  (Read 212 times)

hutch--

  • Administrator
  • Member
  • ******
  • Posts: 8892
  • Mnemonic Driven API Grinder
    • The MASM32 SDK
Anomolies of different PC timings
« on: November 20, 2021, 11:02:54 PM »
I have spent the afternoon setting up the box with an Atermiter Chinese motherboard and with most of it set up and running correctly, I wanted to test if a performance lag behind another Xeon box had improved any. The results show that it is still a reasonable amount slower that the Gigabyte board Xeon box.

The test was done using the identical collection of 4k MP4 files being processed by FFMPEG via a batch file and timed with a recent toy. Each test was run on the local machine with no network data transfer. The i7 clocks close enough to identical to my old i7 so I did not run the test on it, the current i7 was run to test the difference with the faster Xeon box.

About the only difference I could see while the test was running is that the slower Xeon box was running at a lower percentage for all cores where the Gigabyte with the same Xeon was running each core at close to 100%.

The Atermiter board BIOS is close enough to being unintelligible and Xeon CPUs are locked so its not a simple overclock difference, I am guessing that it has some to do with the turbo not working on the Atermiter board BIOS.


BOX         TIME        CPU             Cores       Frequency   X99 Board

Atermiter   17:24   Xeon E5 2690 v3     12 Core     @ 3 gig     Atermiter X99
Hutch       20:34   Haswell i7 5820k     6 Core     @ 4 gig     Gigabyte X99 Gaming 5
Xeon        13:42   Xeon E5 2690 v3     12 Core     @ 3 gig     Gigabyte X99 Gaming 5

hutch at movsd dot com
http://www.masm32.com    :biggrin:  :skrewy:

hutch--

  • Administrator
  • Member
  • ******
  • Posts: 8892
  • Mnemonic Driven API Grinder
    • The MASM32 SDK
Re: Anomolies of different PC timings
« Reply #1 on: November 21, 2021, 12:32:49 AM »
The difference seems to be the BIOS. Found a review from a guy in the Ukraine and basically its a DUD BIOS. Seems that even with an option to enable the turbo boost in the BIOS, the turbo boost does not work.

https://www.youtube.com/watch?v=APFGumn0s1s

I have a new board that I am waiting for some memory and having tested it with some old memory and it looks a lot better, the BIOS even seems to be intelligible.
hutch at movsd dot com
http://www.masm32.com    :biggrin:  :skrewy:

Gunther

  • Member
  • *****
  • Posts: 3830
  • Forgive your enemies, but never forget their names
Re: Anomolies of different PC timings
« Reply #2 on: November 21, 2021, 11:17:49 AM »
Steve,

The difference seems to be the BIOS. Found a review from a guy in the Ukraine and basically its a DUD BIOS. Seems that even with an option to enable the turbo boost in the BIOS, the turbo boost does not work.

what a Diff'rence a BIOS makes. (similar tune by Dinah Washington).  :cool:
Get your facts first, and then you can distort them.

hutch--

  • Administrator
  • Member
  • ******
  • Posts: 8892
  • Mnemonic Driven API Grinder
    • The MASM32 SDK
Re: Anomolies of different PC timings
« Reply #3 on: November 21, 2021, 11:50:44 AM »
He he, I have this bias (not BIOS) that comes from having 3 perfect Gigabyte X99 motherboards that perform with true excellence, have found a source for TPM 2.0 modules (in China) but have so  far had 2 dud boards from China, the first one would not boot at all and had to brawl with the vendor to get a refund.

The Atermiter is the second one and while it runs OK, it is not a performer and is slower with the identical CPU to the Gigabyte board. Number 3 is a brand new Huananzhi board that so far looks good, powered up first time, debug numbers arrived at the right AA value and it actually has an intelligible BIOS. I have a 12 core E5 2690 v3 Xeon in it and if I can get it to run at the right frequency, I have succeeded.

Would have bought another Gigabyte X99 board but they are very hard to get and if you can find one, the prices are in fantasy land. Imported one from the US but it was broken after wasting some days trying to get it to work correctly, reflashed the BIOS to no avail and eventually gave up on it. Vendor told me a sob story about how he was going broke and in the US, it may be the case.

The Haswell era i7's that I am using perform well and are designed to overclock so I am running 2 of them at 4 gig and they are super reliable with decent cooling. There is a 10 core Broadwell CPU that fits a socket 2011-3 board which can be clocked at 4 gig but I am getting tired of shelling out money on the off chance that you may get a performance improvement.
hutch at movsd dot com
http://www.masm32.com    :biggrin:  :skrewy: