pls ask before, there are some exceptions in this rule.
i.e. QY36 = stepping C2, like QXxx
if you get a HPE server with very early BIOS you are lucky, but have to stay with old BIOS until the last days.
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.
i.e. QY36 = stepping C2, like QXxx
if you get a HPE server with very early BIOS you are lucky, but have to stay with old BIOS until the last days.
It seems that Gigabyte considers D0 as 'production' if you read their BIOS release notes. Some of their BIOS screenshots inside the manuals are showing D0 (from 2022 pre-release era). I'm sticking with Gigabyte even though I'm not too happy about their response on cracked RTX GPU cards. Also, Gigabyte LGA 4677 MBs seem to be in short supply at a reasonable price.
I noticed that a lot of screenshots from different China based sellers are identical. They all share the same screenshots. These sellers are somewhat relatated somehow.
The reason I've specified larger cases was because majority of use-cases requiring 7 slots of x16 PCIe (e.g. GPU rendering or ML compute workstation) would have higher power requirements (350Wx7=2.5KW + CPU and other subsystems), therefore case needs space for convenient mount of two power-supply units. Anyhow this non-ES stuff is offtopic here, discussed on other threads I'm not a big fan of.
D0 microcode is marked as PRD state.
also it unclear why ASUS and ASRock keep this MCU in they're W790 boards.
depends on the model. on ASUS W790 Ace & SAGE QYFP,QYFQ,QYFS.... support overclocking
and works with Kingston Renegade Pro DDR5-6000
...memory copatibility is more BIOS thing, should fine if you can run latest BIOS.
Decided to google again for memory related performance tests, I've compared Conjugate Gradient benchmark (sensitive to cache and memory, but not sure where it's bottlenecking between those two) from results pictured here:
CPU/Score (core/threads/Base Ghz)
8380 1P = 20.65 (40/80/2.3)
8280 2P = 28.98 (28/56/2.7 x2)
8490H 1P = 31.24 (60/120/1.9)
8592+ 1P = 35.42 (64/128/1.9)
8380 2P = 40.31 (40/80/2.3 x2)
8490H 2P = 60.42 (60/120/1.9 x2)
8592+ 2P = 70.95 (64/128/1.9 x2)
And it feels like if we check performance adjusted for core counts and memory frequencies - results attributable to better memory (channels, frequencies, DDR4toDDR5) aren't revolutionary different, even arriving to a conclusion what personally to me further investment into neither Sapphire nor Emerald Rapids are not even close to be justifiable at current ebay prices (despite Prod stepping QS and $ being multiples below Intel's advised prices), I mean I have it already for back then affordable price but won't be upgrading in near time to Dual-socket variant.
but seeing as there's recurring talk about buying ES Xeons for ML/GPU workstations, still a valid point I feel
What about the compatibility of: QY08, 8468, ES CPU, 1.9GHZ, 3.3GHz, x 40, 48C/96T?
Would it be compatible with asus W790? I asked the ebay seller and they told me not compatible.
So is this another exception, or maybe the seller is just not sure and to honest?
Current ebay sellers block the TDP of QY08 in their screenshots and only mention it runs on 350W supporting mainboard.
I saw mentions of higher TDP's for some 4677 processor like 385W.
Can anyone confirm the TDP of QY08?
Last edited:
dual QYFS D0 ES2 350W runs on MS73-HB2 (datasheet: officialy supports 2x 270W)
Last edited:
simply yes. ASUS W790 does not block any TDP, it is an OC board.
you see on the CPU-Z shot the seller runs two QY08 on a dual socket board. chn. mostly use Gigabyte MS73 or SM X13DEI.
some D0 ES2 freeze on windows/linux loader:
this solves the issue.
set CPU C-states to enabled(def.=auto) only then next options comes visible.
and set Package C-State to C2(or C0/C1)
please do not get me wrong:
This is important information/discussion, it would just get lost here, no one will find it.
better to move to a specialized thread.
Here's the full shot.the HWInfo screen show a 56C processor, So something is wrong here, weird.
Besides the usual suspects from GIGABYTE MS03-CE0, MS73-HB1 and HB0.
There are now also newer boards, MW83-RP0, and updates like MS03-6L0 and MS73-HB2.
My question is does GIGABYTE maintain compatibility with D0 and other ES stepping that they previously supported also in their newer boards and BIOS?
There are now also newer boards, MW83-RP0, and updates like MS03-6L0 and MS73-HB2.
My question is does GIGABYTE maintain compatibility with D0 and other ES stepping that they previously supported also in their newer boards and BIOS?
MW83-RP0 is a W790 board.
new. rev. 3 brings only out of the box support of Emerald Rapids (BIOS R01,R02)
Another question is about the P-cores and E-Cores. Official intel specs, list different frequencies for the e-cores and p-cores, and for most of the discussed Xeon cpu's here 16 p-cores are stated. While users also often speak about all-core boost/turbo. Not really differentiating, between the different types of cores. Does it mean that the difference is just not that visible?
For the normal 1700 socket intels for some applications, performance gains can be obtained, by disabling the efficiency 'e-cores', in the bios. Giving more power/cooling to the 'p-cores', enabling higher clocks. In an earlier post in this thread (here) , i RolloZ that disabled cores in the ES versions of Ice-lake based Xeons. Did anyone here try such things on the Sapphire Rapids ES Xeons?
For the normal 1700 socket intels for some applications, performance gains can be obtained, by disabling the efficiency 'e-cores', in the bios. Giving more power/cooling to the 'p-cores', enabling higher clocks. In an earlier post in this thread (here) , i RolloZ that disabled cores in the ES versions of Ice-lake based Xeons. Did anyone here try such things on the Sapphire Rapids ES Xeons?