Folding with Xeon E5 v2?

Cool deal, but your last sentence got my attention ----(I have a crane in my car to handle the weight)
Interesting
 
rock solid sine curves
Your homework results are impeccable. If my power goes down for more than a few minutes, I'll have to stop my Folding machines. Although I have a very good backup power generator, it is not good enough to keep the computers' UPSes from switching on and off continuously. Generators with high enough power capacity and inverter technology were just too expensive.

I very much enjoyed reading your generator summary.
 
Your homework results are impeccable. If my power goes down for more than a few minutes, I'll have to stop my Folding machines. Although I have a very good backup power generator, it is not good enough to keep the computers' UPSes from switching on and off continuously. Generators with high enough power capacity and inverter technology were just too expensive.
I very much enjoyed reading your generator summary.
Glad you liked it :)
I do have 2 smaller Honda inverter generators (Eu-10i and EU-30i). They are fantastic, but can't solve the problem, when there is too much load or load changes, the voltage drops below acceptable levels.

Back to topic:
I just received the shipment. Will be fun tonight :)

Left: 2 retail boxes of the SandyBridge based E5-2687W (8c, 3.1GHz, 150W, 34nm)
Right: 2 retail boxes of the IvyBridge based E5-2697v2 (12c, 2.7GHz, 135W, 22nm)
original.jpg


As we can see, Intel's marketing department was pretty innovative with the design of the new retail box:
  1. No more "Intel inside" message. (Heck, what is in my boxes?)
  2. The picture of the chip layout is much smaller (probably representing the 22nm process shrink vs. SandyBridge)
  3. The box has the same size as the previous generation ("See, it is the same 2011 socket")
  4. Intel marketing drove the innovation curve with 3 new pictures. a) The digitalization of our brains IB-Xeons are seemingly much more effective to support, b) the datacenter picture seems to represent the key target market for this CPU and c) a nice - but unreadable - graphics seem to demonstrate that this CPU is much more capable the anything Intel offered before.
  5. On the old box, the terms "INTEL" "XEON" "PROCESSORS" and "E5" are all in bold letters. With the IB Xeons, there seems to be an understanding in the marketing department, that it is much more important that a customer gets an "INTEL" and a "XEON" vs. getting a "E5 processor".
  6. The V2 letters are bigger than the E5-2697. Probably emphazing, that V2 is more important than the underlying CPU. Too bad, there wasn't a E5-2697 V1 CPU around which the new CPU could be compared to.
Isn't it fun to read tea leaves ? :D

More on the performance later today, when I have time getting the 2 CPUs into the system.

cheers,
Andy
 
What do you do that you can afford that monster of a generator, and where can I sign up?
 
Heck, those four CPUs alone probably cost as much as the generator :eek:
The generator is less than the 2 new CPUs.

That's why I called it a "tax return".

The goverrnment purchased this generator with taxpayer money. They didn't use it for 20 years and sold it off with 90% discount. By purchasing it for this lower price, I got some "tax" back :D

Andy
 
Looking forward to your PPD report, and I'm a little curious why did you still buy two E5-2687W (v1) CPUs while E5 v2 series are doing much better on PPD/watt?

Nathan,
With these CPUs (btw, these are retail versions), I am very curious if a E5-2697v2 system can match the energy efficiency of the i4P systems. My working assumption is that the dual socket system should be in the 600-650k ppd range (with 8102-8105 units). The second assumption is a power consumption well below 400 watt, probably closer to 320-350 watt. So, in a best case scenario of 650k ppd and 320 watt, the dual socket system would achieve more than 2000 ppd/watt. With 600k ppd and 350 watt, this would be 1715 ppd/watt. Both numbers would be better than my E5-4650 system (approx 1500 ppd/watt). If it is really along those lines, I consider this impressive for a dual socket system - given the lever of the QRB system favors the 4 socket systems.
 
Looking forward to your PPD report, and I'm a little curious why did you still buy two E5-2687W (v1) CPUs while E5 v2 series are doing much better on PPD/watt?
quickz,
I got the E5-2687W about a year ago and needed them for a different project (setting some records with high speed I/O which were computationally limited).

As I still had the boxes around I played a bit on the "marketing aspects" and shared the picture above :)

Besides using the new CPUs for folding, I will use them to re-evaluate my "balanced system" approach from last year and will check among other things if I can improve my results with the new 12c chips a bit further.

Andy
 
original.jpg


.
.
.
.
.

<snip>
and c) a nice - but unreadable - graphics seem to demonstrate that this CPU is much more capable the anything Intel offered before.

I think this means that the new server CPUs can build better tables and charts in Excel than anything Intel offered before.

:D
 
Guys,
the perf numbers have to wait.

My ASUS Z9-PE-D8 WS board refuses to accept the new CPU. No boot possible with this CPU type. Backgrading to the 2687W solves the issue of boot, but is not what I wanted to achieve.
Upgraded the BIOS to the latest version (5103) which officially supports the IB XEON's. At least the Q-codes change during boot, so the CPU seems to be doing something. But at the end of the the boot process, Q-code 00 is displayed (and no screen output). According to the ASUS manual, Q-code 00 is not used ....

Minimizing the system (1 CPU, 1 DIMM, ...) didn't solve it either.

Guess I have to get in contact with ASUS,
Andy
 
Bummer,

Post Code 00 is what I get if I try to put B0 chips in my Asus Z9PE-D16/2L. I think it means something like it doesn't have a clue how to initialize the chips. That is pretty disappointing that Asus would mess up that badly. Hopefully they have a beta bios they have been using for testing that they can give you that will get the chips running. Perhaps they need to get a new Intel ME version out or something like that, or update something a little deeper than the bios.
 
Bummer,

Post Code 00 is what I get if I try to put B0 chips in my Asus Z9PE-D16/2L. I think it means something like it doesn't have a clue how to initialize the chips. That is pretty disappointing that Asus would mess up that badly. Hopefully they have a beta bios they have been using for testing that they can give you that will get the chips running. Perhaps they need to get a new Intel ME version out or something like that, or update something a little deeper than the bios.

lacking microcode I would assume. waiting on that bios update!
 
Have you tried a complete CMOS clear? Remove the battery and let it sit for a while. Then try to boot one of the new cpus.
 
Have you tried a complete CMOS clear? Remove the battery and let it sit for a while. Then try to boot one of the new cpus.
Hi DooKey.

Did already reset the RTC.
Did a few swaps of the older SB Xeon and the newer IB Xeon.
Its reproducible: Whatever configuration I tried (# of CPUs, # of DIMMs), the SB Xeon boots, the IB Xeon not
Sent a inquiry to Asus Tech Support.

Meanwhile, I have a second board. The Z9PE-D16. I'll upgrade the BIOS to 5105 first and then try my luck over there.
I would have understood potential issue when using ES chips. But these are official retail versions.

Andy
 
Success! :D

Got them to work with the Z9PE-D16 motherboard . Bios = 5105
System is already folding.
 
Somehow I'm very interested in a 2p system; as I need to feed my family I would not be able to get 2 CPUs at the same time (able yes, but my "government" and "finance minister " will not approve :-(
Planning for 2680v2. Can one start with one CPU and put later a second one in the socket ? As there will be some month between is there any risks the two CPUs wouldn't "like" each other ?
As for motherboard: SM, Tyan or Asus ?
How many memory for BA WU is required ?
 
you can start with 1 CPU and upgrade later. can't speak to the CPUs not "liking" each other, but i'm pretty sure that was an older nitpick that CPUs had with Windows, maybe? someone else can speak better to that.

Seems like Asus is actively working on making their boards Ivy compatible, so that might be the safest bet.

Memory speed, do 1333 or 1600, just make sure to do at least 4 dimms per CPU
 
Christian,
I used the boards with one CPU. Will work in Linux and Windows
No problem to upgrade later.
The ASUS Z9PE-D16 works, my Z9PE-D8 WS still does not. The support technician wrote earlier today, that it should. Mine doesnt.
Alternatively look into the SM boards, Movieman at xtremesystems.org has one running and it seems to work well. Plus it is the better one for OC.

Andy
 
im coming back to folding this year, because im building a new render box.

has anyone tried these 26xx v2 with a supermicro X9DAi? i like the look of the barebone system they have with it, because im basically lazy.
I was thinking of buying a single 2680 v2 or 2690 v2 for now and adding another one when i can afford it in a couple of months.
if its going to work, should i just get ram for one CPU now (4x4 im thinking) and get the other ram when i get the other CPU?

and will a single one be able to do bigadv?
 
im coming back to folding this year, because im building a new render box.

has anyone tried these 26xx v2 with a supermicro X9DAi? i like the look of the barebone system they have with it, because im basically lazy.
I was thinking of buying a single 2680 v2 or 2690 v2 for now and adding another one when i can afford it in a couple of months.
if its going to work, should i just get ram for one CPU now (4x4 im thinking) and get the other ram when i get the other CPU?

and will a single one be able to do bigadv?

I'll second that. I am ready to invest in a DP Xeon system but biting off all of the components at once is [H]ard. I was looking at the X9DA7 though. Any thoughts?
Thanks in advance.
 
Supermicro X9DAi supports 26xx v2 with the latest BIOS. A single 2680 v2 is fast enough for doing BA, and only 4 memory sticks are needed for single CPU configurations.
It's should be noted that some PCI-e slots would be unavailable with single CPU configurations on some 2p boards. On Supermicro X9DAi's website it said:
  • [*](Both CPUs need to be installed for full access to PCI-E slots and onboard controllers. See manual block diagram for details.)

has anyone tried these 26xx v2 with a supermicro X9DAi? i like the look of the barebone system they have with it, because im basically lazy.
I was thinking of buying a single 2680 v2 or 2690 v2 for now and adding another one when i can afford it in a couple of months.
if its going to work, should i just get ram for one CPU now (4x4 im thinking) and get the other ram when i get the other CPU?

and will a single one be able to do bigadv?
 
The X9DAI does support the V2 with the R 3.0 BIOS, however if you are using B0 SB-E chips (like I am), you are out of luck with BIOS support.

From Supermicro Tech support:
"The E5-2650 with BO Stepping looks like the Engineering Sample which might not support by our latest BIOS Rev. 3.0. E5-2650 with C1 Stepping has been qualified by our Lab."
 
Yes, B0 stepping SB-E (E5 v1) is not supported by many boards any longer with latest BIOS.
But for the IVB-E (E5 v2), as for as I see, B1 chips (stepping=2) are well supported by X9DAi with BIOS Rev 3.0.

The X9DAI does support the V2 with the R 3.0 BIOS, however if you are using B0 SB-E chips (like I am), you are out of luck with BIOS support.

From Supermicro Tech support:
"The E5-2650 with BO Stepping looks like the Engineering Sample which might not support by our latest BIOS Rev. 3.0. E5-2650 with C1 Stepping has been qualified by our Lab."
 
DeFex said:
im coming back to folding this year, because im building a new render box.

Welcome back!

DeFex said:
I was thinking of buying a single 2680 v2 or 2690 v2 for now and adding another one when i can afford it in a couple of months.

If you're buying one CPU at a time, you might consider the CPUs AndyE just bought (12-core). I would save for the E5-2695 v2 or E5-2697 v2. :D
 
Yes, B0 stepping SB-E (E5 v1) is not supported by many boards any longer with latest BIOS.
But for the IVB-E (E5 v2), as for as I see, B1 chips (stepping=2) are well supported by X9DAi with BIOS Rev 3.0.

Yes but B1/B2 chips very rarely work in a 2p board so that's even worse, also bios support was pulled ages ago in most updates
 
I think this is ONLY true for E5 v1. :)
As for E5 v2, the defination of stepping is different, (as far as I see) B1/B2 stepping is nearly bug free and is supported to work in 2p mode (at least on supermicro 2p boards).

Yes but B1/B2 chips very rarely work in a 2p board so that's even worse, also bios support was pulled ages ago in most updates
 
Sorry for bumping an old thread, but I scored a B0 12c v2 off ebay. Anyone actually got some mobo/BIOS combos that work instead of those that don't? :p
 
Sorry for bumping an old thread, but I scored a B0 12c v2 off ebay. Anyone actually got some mobo/BIOS combos that work instead of those that don't? :p

Sorry been away on hols and just got back, you could try an asus Z9PA-D8. I have my v2 12c in one and they work fine.

A few words of caution. Mine needed memory from the QVL list to work and even then it was a day of various settings to get them to work. Well worth the 500k PPD I can usually get out of them though
 
Back
Top