Ah right. I read about that. Hopefully there will be a V4 exploit in the future 
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.
Greetings,
I'm looking to setup a dual socket low power homelab mainly for VMs
E5-2628L V4 is pretty much what I have in mind and found these steppings on ebay:
QHV8
Intel Xeon Processor E5-2628L v4 ES CPU 1.5GHz 12-Core 30M 65W Low Power QHV8 | eBay
QS QK9B
Intel Xeon E5-2628L V4 QS QK9B 1.9G 12-Core 24-Thread CPU Processor | eBay
Not too sure which server mainboard can I pair the CPUs with - can I even use Xeon ES on dual socket config?
I'm going this route as the Xeon D-1587 out of my budget
I'd appreciate any inputs
Thanks in advance
I'm looking to setup a dual socket low power homelab mainly for VMs
E5-2628L V4 is pretty much what I have in mind and found these steppings on ebay:
QHV8
Intel Xeon Processor E5-2628L v4 ES CPU 1.5GHz 12-Core 30M 65W Low Power QHV8 | eBay
QS QK9B
Intel Xeon E5-2628L V4 QS QK9B 1.9G 12-Core 24-Thread CPU Processor | eBay
Not too sure which server mainboard can I pair the CPUs with - can I even use Xeon ES on dual socket config?
I'm going this route as the Xeon D-1587 out of my budget
I'd appreciate any inputs
Thanks in advance
QK9B is a QS, runs on (dual) asus asrock rack and supermicro mobo, the advantage is the same cpuid(406F1) than the prod.unit.
there is an errata that crashes vm mashines, this is solved by microcode ver 2B an up.
the QHV8 is early stepping(cpuid 406F0) this microcode stucks at rev 14, no more errata will be solved for those cpuid.
pls search also for "Xeon E5 2628 Lv4 QS" gives different results
Last edited:
@RolloZ170 thank you for your input - looks like QS is a better bet
I done more research and found that the E5-2650L V4 has got even higher efficiency/per core for power draw
is the QK93 stepping okay?
Intel Xeon E5 2650L V4 QS QK93 SR2N8 1.7GHz 14Core 35M LGA2011-3 Processor CPU | eBay
this looks promising
Thanks again
I done more research and found that the E5-2650L V4 has got even higher efficiency/per core for power draw
is the QK93 stepping okay?
Intel Xeon E5 2650L V4 QS QK93 SR2N8 1.7GHz 14Core 35M LGA2011-3 Processor CPU | eBay
this looks promising
Thanks again
What do you base efficiency per core on ?
The relationship between a benchmark (which one ?), number of cores and speed, turbo bin's, tdp, t-case temp ?
I am only asking as the whole discussion kind of makes sense sometimes and then you through in t-case temp into the discussion and it goes funny again, e.g. The e5-2628lv4 has a really high t-case temp (or at least higher than the other specs would suggest)
When you talk about efficiency then how does e5-2640v4 vs e5-2680v4 look ? One is a single ring the other a dual ring CPU , how is the base power draw for example ? What effect does dual ring have in a positive or negative way ? The work done per core say based on basic passmark suggests the 2640v4 is much more efficient and also runs cooler at a given TDP but I can't really make that statement as I don't know the specifics of how to rate efficiency.
The relationship between a benchmark (which one ?), number of cores and speed, turbo bin's, tdp, t-case temp ?
I am only asking as the whole discussion kind of makes sense sometimes and then you through in t-case temp into the discussion and it goes funny again, e.g. The e5-2628lv4 has a really high t-case temp (or at least higher than the other specs would suggest)
When you talk about efficiency then how does e5-2640v4 vs e5-2680v4 look ? One is a single ring the other a dual ring CPU , how is the base power draw for example ? What effect does dual ring have in a positive or negative way ? The work done per core say based on basic passmark suggests the 2640v4 is much more efficient and also runs cooler at a given TDP but I can't really make that statement as I don't know the specifics of how to rate efficiency.
efficiency is better at low ghz, from 3 to 4 ghz the drawn power is doubled and so on.......
the more cores the more die in size, more area to cool heat...
t-case is important at passive cooling, if you cool active dont think so much about that.
all frequency at/under 2,5 ghz runs cool.
I am aware of the frequency uses power and generates heat.
I am sceptical when somebody claims X is more efficient than Y as in general for a given core generation they idle similar and the TDP etc is an artificial limit on how much workload a chip can do simply to keep its power consumption and heat within spec.
There is an argument that the high ghz chips are actually more efficient since they are a better bin from the production line they operate with lower power and heat at lower frequency compared to say a TDP limited CPU.
I am sure the internal cpu config, eg number of rings has a difference as scaling changes below 10 core to above 10 core for example.
sure, but this is the wrong place for that
indeed, back to the topic is probably a good idea.People are actively looking for that particular chip too. I originally bought my 2683V3 for 260. When I sold it, I sold my 2695V3 as well. The 2683 sold for 413 and the 2695 sold for 420 (broke even on the 2695 after fees). I found it odd that the sold for only a few bucks more. But yes like you said in a future post (weird) the hack did drive up the price of V3 chips.
I intend to build a home lab based on Asrock EPC612D4I. (LGA2011-3) with Esxi 6 hypervisor
I also need video hardware acceleration for CAD software. - ATI Radeon GPU PCIe card to be passtrought to a windows VM.
I understand that ES processors don't have VT-d capabilities so it will be impossible to do the pci passtrought.
The QS processors have this ability - the VT-d capability?
Or just the OEM genuine ones?
Thanks in advance for your time to answer.
I also need video hardware acceleration for CAD software. - ATI Radeon GPU PCIe card to be passtrought to a windows VM.
I understand that ES processors don't have VT-d capabilities so it will be impossible to do the pci passtrought.
The QS processors have this ability - the VT-d capability?
Or just the OEM genuine ones?
Thanks in advance for your time to answer.
AFAIK BIOS and MoBo must support VT-d too.
some real ES v3 have VT-d.
QS will be useless without VT-d
(if prod.unit supports)You CPU need latest microcode support,because some errata in v3/v4 causes PSOD in Esxi and other VM.
get QS or same stepping than prod.unit and you going fine.
Thank-you for your quick answer. The mobo support VT-d. Now the CPU must support it. How can i be "safe " on buying an QS V3 E5 processor from Ebay. I learned that i may be screwed up. Could you, when you have some little time, advice me in that direction. What should i be aware of...
Thank you again for your patience.
Thank you again for your patience.
If you buy a QS, check if the prod.unit version supports VT-d.
there are SKU's that don't have VT-d.
QS have same base clocks/turbo ratios than prod.units. brand string is complete(not only "genuine intel")
how man cores u r looking for ? base clock ?
aware of NOT QS cpu
remember: even if the seller writes QS, must not be one.
I was intending for 12 to 14 cores max. Probably 12. Base clock: i dont know....sorry, newbie, mobo support DDR4 2133 ECC SO-DIMM, but i think that is a wrong answer.....
OK.
$150...$200 E5-2648Lv3 QS 12C 1.80ghz remarked SR1XW or QFR2
$250...$300 E5-2628LV4 QS 12C 1.90ghz remarked SR2NC or QK9B
Thank you a lot. I will do a deep search and ask the seller to show me a screen of Intel Diagnostic Tool with VT-d: yes.