More tdp = more computing power. It's good option, but most MB's manufacturers recommend a TDP of 270 watts. You need to be careful (if continuous operation is expected)
ES Xeon Discussion
- Thread starter britinpdx
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
32cores seem to be the sweet spot (but Based on ZEN 2 and 3 scalling of the app) then higher clocks/cache/ram bandwith help more then adding core. Dropping a 64core zen3 to 32core actually is faster (thinking power,cache,ram bandwith probably)
this is for many hours or days compute. So yes 24h use. Got this 8352 chip for 80$ and run on the latest bios version. Will look into other 32C intel of LGA4189 variety (somehting like 8358 maiby. Shame the chip 8375C cant run with newer bios)
Thanks for the info
this is for many hours or days compute. So yes 24h use. Got this 8352 chip for 80$ and run on the latest bios version. Will look into other 32C intel of LGA4189 variety (somehting like 8358 maiby. Shame the chip 8375C cant run with newer bios)
Thanks for the info
Last edited:
Hello
We have recently started using a Gigabyte MU72 SU0 with Xeon 4189 (8362B?) ES CPU.
From memory this CPU had a Multi Thread Benchmark >14000 using CPUZ
Please see post 4894 on page 245, Hwinfo64 reports this CPU as a Xeon 8362B SRKYM with a D2 Stepping.
The machine is currently running a task, but will be finished that sometime tomorrow.
FYI this machine is still running Bios F23, I solved the TPM problem by obtaining a CTM010 SPI TPM - the horizontal version.
I will report some further metrics tomorrow.
I hope that helps.
RedX1
We have recently started using a Gigabyte MU72 SU0 with Xeon 4189 (8362B?) ES CPU.
From memory this CPU had a Multi Thread Benchmark >14000 using CPUZ
Please see post 4894 on page 245, Hwinfo64 reports this CPU as a Xeon 8362B SRKYM with a D2 Stepping.
The machine is currently running a task, but will be finished that sometime tomorrow.
FYI this machine is still running Bios F23, I solved the TPM problem by obtaining a CTM010 SPI TPM - the horizontal version.
I will report some further metrics tomorrow.
I hope that helps.
RedX1
Hello
Gigabyte MU72 SU0 with Intel 36 core 000%0 CPU
Thank you for pointing out that this is not QS processor and that HWinfor64 identifies it incorrectly.
Here is the CPU information from the Bios.
ICX D0
Genuine Intel CPU 000%0
CPU Signature - 606A6
Microcode 0D0003A5
Here is the CPUZ Bench for this unit.
Here is the Cinebench R23 Score. 36503pts
During the Cinebench run all cores were ~3.0Mhz and the power at the wall was about 310W.
When this unit is at Idle the power at the wall is 50-60W.
This unit is currently using the F23 Bios,
The latest Gigabyte F40 Bios lists the 606A6 CPU in the release notes.
Will it be dangerous to update the currently used F23 Bios on this unit to the latest available F40 Bios?
I am very impressed with the performance and efficiency of this combination.
Now to find some more appropriate memory at an affordable price.
I hope this helps, and thank you again for all the advice and help.
RedX1
Gigabyte MU72 SU0 with Intel 36 core 000%0 CPU
Thank you for pointing out that this is not QS processor and that HWinfor64 identifies it incorrectly.
Here is the CPU information from the Bios.
ICX D0
Genuine Intel CPU 000%0
CPU Signature - 606A6
Microcode 0D0003A5
Here is the CPUZ Bench for this unit.
Here is the Cinebench R23 Score. 36503pts
During the Cinebench run all cores were ~3.0Mhz and the power at the wall was about 310W.
When this unit is at Idle the power at the wall is 50-60W.
This unit is currently using the F23 Bios,
The latest Gigabyte F40 Bios lists the 606A6 CPU in the release notes.
Will it be dangerous to update the currently used F23 Bios on this unit to the latest available F40 Bios?
I am very impressed with the performance and efficiency of this combination.
Now to find some more appropriate memory at an affordable price.
I hope this helps, and thank you again for all the advice and help.
RedX1
Last edited:
QWAU.
HWinfo uses a table of specs. and if QWAU is unknown by the author the other matching cpu is shown(rarely happen).
in theory he could check the brand string of course, if "0000" = not QS/prod.unit/retail
Hello sir. My friend bought MS73-HB1 and want to use Pmem300. But we found even F06 BIOS had remove the support for Pmem. anyone know which BIOS for this motherboard can use it?
I've had luck modding the Tyan S5652 to run sapphire rapids ES (Q071) and emerald rapids es (Q2SP), and I've been able to mod the Tyan S7130 for sapphire rapids es (Q071) support, but I'm having trouble modding the Tyan S7130 for emerald rapids es (Q30H, A0 stepping) support. It seems that the same basic patch (performed using mmtool) for the microcode that functioned on s5652 for the emerald rapids cpu is not working for the S7130 -- and the culprit is likely that the branch that the S5652 bios (v201-EMR) is based on includes some deeper support for the A0 stepping that they didn't remove, while the oldest available EMR-compatible for the S7130 (v202) might not have the required code -- it gets stuck on iio init with code FF10/FF05. I've got a few questions as I work on modding this board:
1. Has anyone already modded this board for A0 support? -- this would simplify things greatly
2. Does anyone have a bios revision that may be from a different branch for this board that's not available online? I've scoured their downloads and found nothing, including for related boards like the s7136.
3. Does anyone have the A0 patched bios for the gigabyte MS73-HBx? -- this is the hail-mary here, where I hope to analyze the modules to see if I can extract anything to patch into the S7130 bios
Edit: this is the bios that I am trying to mod: https://www.mitaccomputing.com/en/downloads/details/MitacServer/S7130GM2NR-2T_BIOS_4591
1. Has anyone already modded this board for A0 support? -- this would simplify things greatly
2. Does anyone have a bios revision that may be from a different branch for this board that's not available online? I've scoured their downloads and found nothing, including for related boards like the s7136.
3. Does anyone have the A0 patched bios for the gigabyte MS73-HBx? -- this is the hail-mary here, where I hope to analyze the modules to see if I can extract anything to patch into the S7130 bios
Edit: this is the bios that I am trying to mod: https://www.mitaccomputing.com/en/downloads/details/MitacServer/S7130GM2NR-2T_BIOS_4591
Last edited:
It works! Thx so much! (I used the hpm) -- so far it's on memory channel trainingtry this mod. BIOS. i don't have S7130 so i can't test.
Dropbox
www.dropbox.com
edit: boots, 5600MT/s RAM
Attachments
Last edited:
Fantastic thread and resource for the community--thank you all in advance!
I have two 8368 QWATs that I am trying to get working in a Supermicro X12DPG-QT6 motherboard. Each CPU works fine by itself when installed in the CPU1 socket and the system POSTs without error. But as soon as I add the second CPU to the CPU2 socket, the system tries to POST but then shuts down after a few seconds with the BMC health event log showing errors: "CPLD [CPU-9008] CPU1 THERMAL TRIP - Assertion" and "CPLD [CPU-9009] CPU2 THERMAL TRIP - Assertion". Neither CPU is overheating because they work fine by themselves so the error is very mysterious and unhelpful.
I get these same errors regardless of the DIMM population (I've tried with just one DIMM in the A1 position of each socket), and I have removed all peripheral components and the motherboard is sitting on a cardboard box. I'm stumped. Appreciate any and all wisdom / advice / modified BIOS or firmware / ancient evil rituals / etc., that you all think might help.
I have two 8368 QWATs that I am trying to get working in a Supermicro X12DPG-QT6 motherboard. Each CPU works fine by itself when installed in the CPU1 socket and the system POSTs without error. But as soon as I add the second CPU to the CPU2 socket, the system tries to POST but then shuts down after a few seconds with the BMC health event log showing errors: "CPLD [CPU-9008] CPU1 THERMAL TRIP - Assertion" and "CPLD [CPU-9009] CPU2 THERMAL TRIP - Assertion". Neither CPU is overheating because they work fine by themselves so the error is very mysterious and unhelpful.
I get these same errors regardless of the DIMM population (I've tried with just one DIMM in the A1 position of each socket), and I have removed all peripheral components and the motherboard is sitting on a cardboard box. I'm stumped. Appreciate any and all wisdom / advice / modified BIOS or firmware / ancient evil rituals / etc., that you all think might help.
make sure all 2x 4 T30 screws are fasten to the end(crosswise,one turn per stroke), torque-screwdriver is helpfull but not required.
try oldest possible BIOS to sort that out.
from H12/X12 and up Supermicro uses Root of Trust Firmware, even after change of one bit it needs resign with supermicro private key/cert.
brand new motherboard ? it is a known issue the boards bricks if you try one CPU only in second socket.
No--used from ebay. But would it still POST without issue from the first socket if the previous owner had tried to boot with only the second socket and damaged it?
I will try the oldest BIOS I can find next--looks like 1.1 on https://drunkencat.net/misc/SupermicroBIOS.html.
Do you think there is anything to it being a CPLD error? I am planning on using optane PMEM DIMMs if I can get it to POST with both CPUs and I think I read somewhere in this thread that PMEM can require a CPLD update, which I guess I would have to request from Supermicro?