Dual E5645 mini-build log

[twilyth]

If it's not coming up in BIOS, that' gonna be some type of Hardware prob, as opposed to software. Start with checking all the power connectors are properly flush, then from BIOS load optimized defaults and reset CMOS (page 2-24 of manual)

Hope it ain't a bad CPU. Certainly 2P boards are a whole lot more hassle than we're used to.

Ah. OK. Good point. I'll take a photograph of the bios screens

 

[KieX]

How the fuck do they call it a server? Aw, nevermind. I get it. HOME server, not server server.

OK. I'll load w7 pro then. Hope I can do an upgrade install.

Here is the CPUZ display. If the the manual lists the sockets as 1 and 2, I assume that CPUZ means socket 0 when it says #1?

Thanks very much man.

Just saw his post after mine lol.

Perhaps Server2008 is in order?
http://www.microsoft.com/download/e...b00000c932fd0rjc7_5p3t&displaylang=en&id=5023

 

[stinger608]

KieX could be correct. Not sure why it would not be showing in the bios to be honest. that really shouldn't be affected from the operating system. Hmmm, very odd indeed.

 

[twilyth]

Just saw his post after mine lol.

Perhaps Server2008 is in order?
http://www.microsoft.com/download/e...b00000c932fd0rjc7_5p3t&displaylang=en&id=5023

OK, but if you all could take a look at my screen shots first and let me know what you think.

 

[stinger608]

I don't know, its really not showing either a single chip or 2 chips. I think it is because both chips are identical. I think I would load up Windows 7 Professional and see if the second physical chip is seen in Windows.

 

[KieX]

Think you may need to go back a couple of steps. Advanced> Hardware Monitor

It should show you temps and such for both CPU's. If it's not.. then yeah HW problem.

 

[bogmali]

CPU-Z somehow is showing your E5645's as E5620's? Last time I checked E5620's are quad Westmeres and not hexes. So did you get CPU0 and CPU1 to show up?

 

[twilyth]

Think you may need to go back a couple of steps. Advanced> Hardware Monitor

It should show you temps and such for both CPU's. If it's not.. then yeah HW problem.

It looks like it is seeing both, but one is very hot and has no rpm on fan speed. Will double check connections, but both fans are running.

EDIT: dling server 2008 R2 SP1

 

[stinger608]

Looks like you have one of the cpu fans plugged into a chassis fan header man. But yea, both physical processors are showing up just fine in the bios.

Server 2008 R2 will cure the issue in windows

 

[twilyth]

Looks like you have one of the cpu fans plugged into a chassis fan header man. But yea, both physical processors are showing up just fine in the bios.

Server 2008 R2 will cure the issue in windows

Thanks again. Burning the disk now. Will power down, double check headers, and install. I did check the headers the first time but I will double check with the manual now.

 

[twilyth]

UPDATE: pulled off the cooler on cpu 2, de-spooged and reapplied paste - this time with Noctua (couldn't find it the first time around). Fired it up and the temps were identical.

The first cpu is right next to a case fan so I think I'm getting better temps with the Noctua paste but the extra airflow is helping cpu 1.

Anyway, halfway through the install. Have to run an errand and will re-install boinc when I get back in about an hour.

Oh, and it turns out I was hallucinating. I did indeed have it on the wrong header. I could have sworn I checked it though. You see, this is why I have OCD. I really can't trust myself.

edit: have to take a shower, so I guess I'll go out later.

 

[repman244]

Fired it up and the temps were identical.

Don't worry much about having different temperatures since the chips don't necessarily have the same V core. As I can see from the picture your CPU 1 is running at 1.212V and CPU 2 at 1.218V.
My 2 old Xeons also have very different V core and run at about 10C different.

 

[twilyth]

@Repman - thanks buddy!

With Win server 2008 installed I now have 24 glorious threads. I can almost hear the angels singing!!!

That one angel looks a little gay, but as long as he can crunch . . .

Thanks everyone for your help. I never would have figured this out on my own. My team rocks!!!

 

[gplnpsb]

I didn't want to clutter up the team page and this isn't going to be a big deal with any mods or such. It's just mainly going to be a pix gallery. BTW, it seems that I bought some engineering samples rather than honest to god E5645's. Anyone want to very what I have? I'll look at the imprints tomorrow but couldn't really make it out tonight.

Q2WV's are A-0 stepping engineering samples that are basically equivalent to the B-1 stepping mass production E5645s. I should note that they have locked multipliers, as they have a dual-qpi link. Only the single qpi link gulftown engineering samples (ie Q3FE) had unlocked multipliers. If it interests you, your chip's batch number C9300170 indicates that it was manufactured in the 30th week of 2009, basically the last week of July.

CPU-Z somehow is showing your E5645's as E5620's? Last time I checked E5620's are quad Westmeres and not hexes. So did you get CPU0 and CPU1 to show up?

CPU-Z sometimes has issues generating the name string for engineering samples that are from before the qualification sample stage. The name string is basically an estimate of what the cpu should be based on the cpuid values it reports. The specification string is the one to trust, its an ASCII string reported directly by the cpu. Here it just reads "Genuine Intel(R) CPU", indicating that this is a pre qualification sample ES that was manufactured before the Gulftown chip specifications and branding were finalized.

 

[twilyth]

@gplnpsb: Thanks man. Interesting stuff. Not to be argumentative, but the ASUS board did let me kick the multi from 18 to 19 and get a final clock of >2.5ghz. But either way, it's cool. I'm just happy to have the damned thing working. I feel like such an ass too because I'd already contacted the seller, but I just sent him an email, so that's settled now too. Miller time.

btw, if you think you might want to contribute some cpu cycles to the cause, we'd be happy to have you, but either way, welcome to TPU!

 

[mstenholm]

Just saw his post after mine lol.

Perhaps Server2008 is in order?
http://www.microsoft.com/download/e...b00000c932fd0rjc7_5p3t&displaylang=en&id=5023

I used that version when I had a 4P box. It works.

 

[Completely Bonkers]

24 thread! What The Funk! Nice rig man. Makes me want to upgrade my SLOW dual E5420s.

How about a doing some benchies for us to drool over
Like Geekbench
Or cinebench R10 or R11.5 / CPU parts
Or gflops bench
Or Fritz

 

[twilyth]

24 thread! What The Funk! Nice rig man. Makes me want to upgrade my SLOW dual E5420s.

How about a doing some benchies for us to drool over
Like Geekbench
Or cinebench R10 or R11.5 / CPU parts
Or gflops bench
Or Fritz

Ahem. So . . . let me see if I understand. You want me to use some of my precious CPU cycles to do benchies.

Hmmm. Why don't you ask my buddy Smeagol how we feel about the precious.

j/k. I AM a points whore but not THAT bad - yet.

I have to go out (never made it before), but I'll dl those guys now.

 

[Completely Bonkers]

I find running benchmarks very boring too! And a few % here and there isnt important. But seeing how much your new rig ZOMGS my E5420 workstation might encourage me to do something about it...! I'm either going to upgrade the CPUs for an extra 15-20%. Or I'll junk it and get a whole new rig.

 

[twilyth]

I find running benchmarks very boring too! And a few % here and there isnt important. But seeing how much your new rig ZOMGS my E5420 workstation might encourage me to do something about it...! I'm either going to upgrade the CPUs for an extra 15-20%. Or I'll junk it and get a whole new rig.

I understand, I was just busting your chops.

It seems that I should probably keep my sense of humor to myself. I had posted something to the team thread that didn't go over very well either. I had trouble pronouncing Shistosoma so I wanted to change it to Snoo Snoo. It probably didn't help that it made no sense whatsoever, but that was sorta the point. Oh well . . .

I'll definitely run everything I can get for free in terms of benchies. Cinebench doesn't seem to have a free version. Intel burn test is a definite. Not familiar with the other 2 but I have them on a thumb drive ready to go.

 

[twilyth]

I find running benchmarks very boring too! And a few % here and there isnt important. But seeing how much your new rig ZOMGS my E5420 workstation might encourage me to do something about it...! I'm either going to upgrade the CPUs for an extra 15-20%. Or I'll junk it and get a whole new rig.

I couldn't get burntest to run.

fritz chess would only use 8 cores

I was able to run geek bench, but the free version only allows 32-bit mode. Results are attached.

Benchmark Summary
  Integer Score             17482 |||||||||
  Floating Point Score      20675 |||||||||||
  Memory Score               4186 ||
  Stream Score               5062 ||

  Geekbench Score           14698 |||||||

System Information
  Operating System      Microsoft Windows Server 2008 R2 Standard (64-bit)
  Model                 System manufacturer System Product Name
  Motherboard           ASUSTeK Computer INC. Z8NA-D6(C)
  Processor             Genuine Intel(R) CPU             000  @ 2.40GHz @ 2.41 GHz
                        2 Processors, 12 Cores, 24 Threads
  Processor ID          GenuineIntel Family 6 Model 44 Stepping 0
  Processor Package     Socket 1366 LGA
  Processor Codename    Westmere-EP
  L1 Instruction Cache  32.0 KB x 6
  L1 Data Cache         32.0 KB x 6
  L2 Cache              256 KB x 6
  L3 Cache              12.0 MB
  Memory                24.0 GB DDR3 SDRAM 667MHz
  Northbridge           Intel 5500 22
  Southbridge           Intel 82801JR (ICH10R) 00
  BIOS                  American Megatrends Inc. 1209   

Integer Performance
  Blowfish
    single-threaded scalar   1451 
    multi-threaded scalar   29785 ||||||||||||||||
  Text Compress
    single-threaded scalar   1897 |
    multi-threaded scalar   30001 ||||||||||||||||
  Text Decompress
    single-threaded scalar   2043 |
    multi-threaded scalar   28783 |||||||||||||||
  Image Compress
    single-threaded scalar   1922 |
    multi-threaded scalar   31426 |||||||||||||||||
  Image Decompress
    single-threaded scalar   2047 |
    multi-threaded scalar   26300 ||||||||||||||
  Lua
    single-threaded scalar   3400 |
    multi-threaded scalar   50732 |||||||||||||||||||||||||||

Floating Point Performance
  Mandelbrot
    single-threaded scalar   1904 |
    multi-threaded scalar   36286 |||||||||||||||||||
  Dot Product
    single-threaded scalar   1048 
    multi-threaded scalar   22933 ||||||||||||
    single-threaded vector   4707 ||
    multi-threaded vector   73506 ||||||||||||||||||||||||||||||||||||||||
  LU Decomposition
    single-threaded scalar   2360 |
    multi-threaded scalar    7672 ||||
  Primality Test
    single-threaded scalar   3903 ||
    multi-threaded scalar   43000 |||||||||||||||||||||||
  Sharpen Image
    single-threaded scalar   1447 
    multi-threaded scalar   28433 |||||||||||||||
  Blur Image
    single-threaded scalar   2935 |
    multi-threaded scalar   59317 ||||||||||||||||||||||||||||||||

Memory Performance
  Read Sequential
    single-threaded scalar   4673 ||
  Write Sequential
    single-threaded scalar   6511 |||
  Stdlib Allocate
    single-threaded scalar   2954 |
  Stdlib Write
    single-threaded scalar   3030 |
  Stdlib Copy
    single-threaded scalar   3764 ||

Stream Performance
  Stream Copy
    single-threaded scalar   4752 ||
    single-threaded vector   6015 |||
  Stream Scale
    single-threaded scalar   4330 ||
    single-threaded vector   5652 |||
  Stream Add
    single-threaded scalar   4677 ||
    single-threaded vector   5800 |||
  Stream Triad
    single-threaded scalar   4910 ||
    single-threaded vector   4360 ||

 

[stinger608]

That is frigging awesome Twilyth!!!!!!!!!!

Finally up and crunching 24 glorious threads.


Oh and I just sent you a PM bro. Regarding something of interest for this system.

 

[mstenholm]

Finally up and crunching 24 glorious threads

I just saw the first real numbers from your new build at Free-DC and I noticed that yours "bench" higher then Kiexs ditto 24 threads in Floating Point Speed: 2,286.91 vs 2,062.86 million ops/sec. Only time will show...

Talking about time then I'm 3-3½ day away from my blue SNtS goal and that coincides with a trip to Africa (not a holiday destination sadly) so I will be down to around half my current production.

Edit: sorry Kiex, more like 2 days from now

 

[twilyth]

I just saw the first real numbers from your new build at Free-DC and I noticed that yours "bench" higher then Kiexs ditto 24 threads in Floating Point Speed: 2,286.91 vs 2,062.86 million ops/sec. Only time will show...

Talking about time then I'm 3-3½ day away from my blue SNtS goal and that coincides with a trip to Africa (not a holiday destination sadly) so I will be down to around half my current production.

Edit: sorry Kiex, more like 2 days from now

Wow. I didn't even know we could still find our bench stats anywhere. Thanks. That 's a very cool feature of Free-DC.

But I just checked and I think you may have been looking at a different rig.

for 1921802 it says Floating Point Speed: 2,315.49 million ops/sec

He also has 1696780 which says Floating Point Speed: 2,062.86 million ops/sec

mine is still what you quoted for rig 1921792

but it's also listed as 1921320 which says Floating Point Speed: 2,254.05 million ops/sec

I think we both did multiple installs so in my case, those 2 rigs are the same. Not sure about Kiex but I know he did a lot of tweaking over the weekend.

But regardless - thanks again man.

 

[KieX]

I just saw the first real numbers from your new build at Free-DC and I noticed that yours "bench" higher then Kiexs ditto 24 threads in Floating Point Speed: 2,286.91 vs 2,062.86 million ops/sec. Only time will show...

Talking about time then I'm 3-3½ day away from my blue SNtS goal and that coincides with a trip to Africa (not a holiday destination sadly) so I will be down to around half my current production.

Edit: sorry Kiex, more like 2 days from now

Wow. I didn't even know we could still find our bench stats anywhere. Thanks. That 's a very cool feature of Free-DC.

But I just checked and I think you may have been looking at a different rig.

for 1921802 it says Floating Point Speed: 2,315.49 million ops/sec

He also has 1696780 which says Floating Point Speed: 2,062.86 million ops/sec

mine is still what you quoted for rig 1921792

but it's also listed as 1921320 which says Floating Point Speed: 2,254.05 million ops/sec

I think we both did multiple installs so in my case, those 2 rigs are the same. Not sure about Kiex but I know he did a lot of tweaking over the weekend.

But regardless - thanks again man.

Yeah, sorry about that. The current 2.57GHz is the 2315 million ops/s

I didn't have a spare PSU over weekend, so I took it from another cruncher for the SR-2 and copied the BOINC data folder to finish off those WU. Got the PSU through post today, so the previous entry will go back to being a 2600K, and the newly finished SR-2 will take a few days to ramp up from the bottom of my list.

Try to enjoy your trip MStenholm, should come visit here in London sometime