even a with Xeon W-2400 the motherboard will power on but not POST.
defective CPU ?
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.
defective CPU ?
I'm running a MS73-HB1 and 2xQYFQ, and I haven't been able to get NVMe drives to work in the m.2 slots on the motherboard, they aren't recognised in the bios. I've tried an HP FX900 and a Samsung 970 evo. I'm using R07 bios (it didn't work on the 2023 factory bios either).
Using a PCIe to NVMe adapter works fine. Is there a setting I need to enable somewhere to get the slots on the motherboard working?
Using a PCIe to NVMe adapter works fine. Is there a setting I need to enable somewhere to get the slots on the motherboard working?
Hi everyone,
I’m running dual Xeon 8260 ES QQ89 processors on an Intel S2600WF motherboard.
The system boots fine, and all cores are correctly detected. Initially, it was very unstable, but after reading about the C-state issue, I disabled it, and now the system is stable.
However, I noticed there's no microcode in the BIOS.
Using lscpu, all features, including AVX-512, are detected and working. I also tested with Geekbench 6.
That said, Geekbench 6 identifies the CPUs as “Intel Pentium III,” and the CPU score is much lower than expected.
Another issue is the high idle power consumption—around 140W!
The microcode (50655) is missing from the BIOS. I managed to extract the correct microcode from a Supermicro BIOS file.
Has anyone successfully modified the BIOS for the S2600WF motherboard to include the missing microcode?
Thanks in advance!
I’m running dual Xeon 8260 ES QQ89 processors on an Intel S2600WF motherboard.
The system boots fine, and all cores are correctly detected. Initially, it was very unstable, but after reading about the C-state issue, I disabled it, and now the system is stable.
However, I noticed there's no microcode in the BIOS.
Using lscpu, all features, including AVX-512, are detected and working. I also tested with Geekbench 6.
That said, Geekbench 6 identifies the CPUs as “Intel Pentium III,” and the CPU score is much lower than expected.
Another issue is the high idle power consumption—around 140W!
The microcode (50655) is missing from the BIOS. I managed to extract the correct microcode from a Supermicro BIOS file.
Has anyone successfully modified the BIOS for the S2600WF motherboard to include the missing microcode?
Thanks in advance!
intel removed A0 stepping. i wonder it still works.
BIOS is signed, Intel BIOS guard...
maybe use SPI flasher, read, modify and flash back,
i was not succeccfull with this, but gave not much time.
but linux and windows(KBxxxxnnnn) are able to update microcode.
I tried to load microcode in Linux, but no success.
bios dump via flashrom. Got a 64 MB file. i can open it in UEFITool NE and check structure.
i did a mod biod file and uploaded from BMC , i writes the bios , and i only did change the microcode.
Just learning about the fit table, need more time to make it work.
Looking at other projects like
LinuxBoot / S2600
i have spi hardware just in case. BMC flash haves allready saved me one time
Spotted QYG3, one I haven't seen mentioned in the thread before: Intel Xeon Platinum 8470 ES CPU LGA4677 2.0-2.9GHz Processors Support MS33-AR0 | eBay
Seller calls it an 8470 but doesn't provide a full picture of the CPU data for comparison.
Seller calls it an 8470 but doesn't provide a full picture of the CPU data for comparison.
BIOS guard is online self check. if the microcode is not part of protected area it should work with SPI flasher.
signature is checked by flash update tool. (protect against modification by third party)
late E0 stepping and up.
E0 e.g. Q0KG,Q0KL,Q0KH,Q0KJ,Q0KS,Q0KM...(cheap at taobao/idlefish/goofish)
edit: all QS work.
Last edited:
I'm trying to do more research on the Sapphire Rapids platform. As I understand it, they have cores that Intel calls High Priority and Low Priority cores rather than Performance and Efficiency cores. Looking at the 4th Gen Xeon CPUs in the comparison on Intel's website, all of the E1A scalables I checked top out at 16 High Priority cores.
Are P-cores the same as High Priority cores but with a different name?
If I disabled all but the 16 High Priority cores, would I be able to hit a higher speed than if I left the Low Priority scores on? Do the D0s even have that distinction in their stage of development?
Are P-cores the same as High Priority cores but with a different name?
If I disabled all but the 16 High Priority cores, would I be able to hit a higher speed than if I left the Low Priority scores on? Do the D0s even have that distinction in their stage of development?
Priority cores are about frequency: What Does It Mean When a Processor Has High-Priority Cores and...
P or E cores about microarchitecture and it means that you have one microarchitecture that is smaller and slower than other one. In that generation Golden Cove was a P core microarchitecture and all cores on Sapphire Rapids are P cores.
yes. but High and Low Priority cores are the same core type.
remaining cores can go higher clocks.
e.g. QYFS D0 has turbo of 3700mhz (1..28C)
disable 28 cores and the remaining 28 cores go 3700mhz.
example: 32C with 16C enabled
Last edited:
I had read that before and it was where I got this idea, not realizing that it wasn't a mix of cores but based on amount of them.
Considering that I wouldn't expect to have more than 24 or 26 cores on an Asrock W790, that sounds very good.
Is there a chart, list, or projected graph anywhere with correlated core counts to clock speed for D0?
intel doesn't provide this for ES, and for prod.unit is under NDA (or just hard to find LOL)
i usualy install CPU and run HWinfo.
Something like this or this (SST-PP section aka Speed Step Technology for some SKUs configurable via BIOS where you can select number of cores from very limited set of configurations) ?
I think it's not widely available for ES versions apart from generic info like in RolloZ170' post from HWINFO64 above (which IMV is different from SST-PP since it's more related to Turbo limits which is normally restricted by duration, although you can select core affinity on operating system level too)
Last edited:
S2600WF BIOS Mod Update
After dumping the BIOS via SPI, I added the updated microcode—surprisingly straightforward for a first attempt!
Finally, no more error messages in the management interface. Previously, the errors were severe enough to halt the system, but now it’s stable.
Turbo Boost seems to work fine. All cores boost to around 3 GHz, and with fewer cores active, it can clock even higher.
Any recommendations for additional tweaks or features to include in a modded BIOS?
I can’t run Geekbench at the moment since it doesn’t support ES chips, but I’ve been using PassMark PerformanceTest for Linux instead: PassMark PerformanceTest.
During benchmarks with all cores turbo-enabled, temperatures stay below 66°C, which is great. Fan control through the BIOS works well, keeping the server relatively quiet.
Performance is comparable to a Xeon 8260, but since this is a dual-CPU system, the benchmark results are roughly double those of a single CPU setup.
I’m still unsure about memory configuration and benchmarking. MemTest86 runs without errors and completed a full pass, but I could use advice on how to properly configure or test the memory for optimal performance.
After dumping the BIOS via SPI, I added the updated microcode—surprisingly straightforward for a first attempt!
Finally, no more error messages in the management interface. Previously, the errors were severe enough to halt the system, but now it’s stable.
Turbo Boost seems to work fine. All cores boost to around 3 GHz, and with fewer cores active, it can clock even higher.
Any recommendations for additional tweaks or features to include in a modded BIOS?
I can’t run Geekbench at the moment since it doesn’t support ES chips, but I’ve been using PassMark PerformanceTest for Linux instead: PassMark PerformanceTest.
During benchmarks with all cores turbo-enabled, temperatures stay below 66°C, which is great. Fan control through the BIOS works well, keeping the server relatively quiet.
Performance is comparable to a Xeon 8260, but since this is a dual-CPU system, the benchmark results are roughly double those of a single CPU setup.
I’m still unsure about memory configuration and benchmarking. MemTest86 runs without errors and completed a full pass, but I could use advice on how to properly configure or test the memory for optimal performance.
