Intel Xeon 5600 Secures, Speeds Cloud Computing

[onex]

The X series Xeon's are only worth the price if used in dual socket mobo's. The W series Xeons are the i7 counterparts (single QPI) and the w3680 is listed by intel as $999; same as the 980x.

the W's should be better OC'ers too generally, VID range drops from 1.375 to 1.3,
though memory bandwidth could become a bit of a limiting factor,
a 4GHz 980 (4*12 full load) vs exp. PC12800 * 3 channel = 48GHz vs 38,400..,

you might need a 2000MB/s or at least highly OC'ed 1800 modules to take care of these extra 4 cores.

why the hell did they make these chips support only 1066 and 25GB/s bandwidth .

E:
anyway, it is shockingly amazing!!
http://www.overclock3d.net/reviews/c...80x_gulftown/6

[trn]

I think Intel stated somewhere they kept the minimum ram support at 1066 to make the chips backward compatible with the current minimum x58 spec.

[onex]

well, maybe it's not entirely linear, cpu has to cycle instructions for every byte or dword taken from memory,
980 should be running at 150,000MIPS - ~45IPS per Hz running 3.33,

i'd love to figure out how it works ,

anyone ?

[clo007]

I've been trying to get this across to everyone for the last 4 years.
Both of them make excellent products.
Sure at any one time one will be better than another but that is due to timing in R&D and who hits the market first.
Same with video cards and everything else in life.
What frosts my balls is the perceived need in some people and the companies themselves to feel they have to dump on the competition.
There's more to say on this and hopefully won't need to be said in the open forum. I just hope the people involved can take a hint before it gets to an all out war.

I say give the bashers/trolls a week off to think about it.

[onex]

anyway, it is shockingly amazing!!
http://www.overclock3d.net/reviews/c...80x_gulftown/6

pff,
last time to trust any result based on a single review.

Oh that? That's just a 47% peformance boost over the Core i7 975. This is nearly perfect scaling with core count, showing us that we're not limited by memory bandwidth or anything else.

1333 7-7-7-20 & (should be) HD5870, he says, memory bandwidth doesn't have any impact on the gulftown performance..


Cinebench R10 1 core.


multi core..


Tweaktown, HD5870 & 4GB Kingston KHX12800D3T1K3 @1600
*all tests with various SSD's X25-M or Kingston SSDnow*

~2.5% faster..
results could be negligible though

[kemo]

any word on Xeon E5620 availability ??

[trn]

Any word on availability for any of these chips? When are they suppose to be available at retailers?

[Clairvoyant129]

yeah right adding ln2 unstable platforms, hardly able to run anything and compare it against a 24/7 platform real blue fanboy.

Can someone run just a dual gulf @ 3ghz and 3.33 do the prime all over again movieman?

So if you overclock, it's considered being a fanboy? Hahaha. AMD guys always give me a good laugh. Don't you think you're at the the wrong site?

Oh and look again, overclocking on air @ 4.1GHz is enough to beat 24C Opterons.

[Movieman]

pff,
last time to trust any result based on a single review.

Here's a POV that you can trust...Mine.
And thats at the same speed I run 100% loaded 24/7 on WCG..
Literally shut down WCG and ran this. No tweaking, no games..
Now where's my good buddy duploxxx?

[Clairvoyant129]

Here's a POV that you can trust...Mine.
And thats at the same speed I run 100% loaded 24/7 on WCG..
Literally shut down WCG and ran this. No tweaking, no games..
Now where's my good buddy duploxxx?

Don't expect anything, every time you challenge him, he mysteriously disappears for a several days.

[Movieman]

Don't expect anything, every time you challenge him, he mysteriously disappears for a several days.

Nah, lets be nice. He is a fairly smart guy just has caught that disease thats been going around this forum for a long time. Fanboyitis..
Comes in two variations and both deadly.
There's the green strain and the blue strain..

[duploxxx]

Don't expect anything, every time you challenge him, he mysteriously disappears for a several days.

Nah, lets be nice. He is a fairly smart guy just has caught that disease thats been going around this forum for a long time. Fanboyitis..
Comes in two variations and both deadly.
There's the green strain and the blue strain..

thx for the credit, appreciated. What are you platform specs HW / OS so we know what we are comparing and I can try to leverage the test?

what settings dit you use, you do know that you are on an old beta there is already a 36.

[Movieman]

thx for the credit, appreciated. What are you platform specs HW / OS so we know what we are comparing and I can try to leverage the test?

what settings dit you use, you do know that you are on an old beta there is already a 36.

Platform..
2001 Pentium P3-1000......

OOP's sorry..
EVGA W555 board
2-X5680 Westmere's at 4352mhz on air
6x2 gig Corsair GTX1's @1780 Cas7
Win7 64 Ultimate
Just did that test a few days back and downloaded from their site.
No changes to the program made.
Just looked for the benchmark in it and ran it.

[duploxxx]

Platform..
2001 Pentium P3-1000......

OOP's sorry..
EVGA W555 board
2-X5680 Westmere's at 4352mhz on air
6x2 gig Corsair GTX1's @1780 Cas7
Win7 64 Ultimate
Just did that test a few days back and downloaded from their site.
No changes to the program made.
Just looked for the benchmark in it and ran it.

then what is the added value of comparing the results when you do your test on a océd and optimized memory platform against a standard platform not even SE version.

the only thing i am interested in is how some x5670,E5645,E5630 go against AMD competition.

and finally what pov are you running?

[Movieman]

then what is the added value of comparing the results when you do your test on a océd and optimized memory platform against a standard platform not even SE version.

the only thing i am interested in is how some x5670,E5645,E5630 go against AMD competition.

and finally what pov are you running?

Like everyone here I test on what I own.
Version is 3.73 as thats what they were sending last week when I tested.
What it showed me is my system scaled pretty damned close to twice the shown results using the i7-980 and then factor in some for my 4352 speed vs the stock 3330.
The real point is that what I showed are attainable numbers as that test was done at trhe same settings I run WCG at and WCG put's a pretty healthy load on the machine with 24 threads.
Just a tad below Prime 95 in the load factor.

[duploxxx]

Like everyone here I test on what I own.
Version is 3.73 as thats what they were sending last week when I tested.
What it showed me is my system scaled pretty damned close to twice the shown results using the i7-980 and then factor in some for my 4352 speed vs the stock 3330.
The real point is that what I showed are attainable numbers as that test was done at trhe same settings I run WCG at and WCG put's a pretty healthy load on the machine with 24 threads.
Just a tad below Prime 95 in the load factor.

whatever, as I stated before, not interested in a compare like that but I'll show you a bit, perhaps If you have some time you could actually do some more testing on normal speed and mem configuration with single sock/dual sock , HT on and off to make a nice comparison.

anyhow not able to provide more details except for the fact that this is a 12core (no SE) with 32GB 1333 mem on win2008 64bit

[OhNoes!]

Where's Movieman when you need him?

[Movieman]

whatever, as I stated before, not interested in a compare like that but I'll show you a bit, perhaps If you have some time you could actually do some more testing on normal speed and mem configuration with single sock/dual sock , HT on and off to make a nice comparison.

anyhow not able to provide more details except for the fact that this is a 12core (no SE) with 32GB 1333 mem on win2008 64bit

6926.. Excellent.. is that a dual 6 core or single 12 core?

[duploxxx]

6926.. Excellent.. is that a dual 6 core or single 12 core?

single

[onex]

12core (no SE)

that's "standart edition" or you rather meant ES?
anyhow, if u couldn't provide any more details, that should be an ES, and if that is THAT ES, then these are fine score .

MM

Here's a POV that you can trust...Mine.

the test above was meant to see whether the 25.6Ghz mem bandwidth for the 980 is actually limiting it,
the results shows a 2.5% difference in favor to the 1600Mhz vs the 1333.
that could be negligible though.
anyway,
show some Cinebench!

[Movieman]

that's "standart edition" or you rather meant ES?
anyhow, if u couldn't provide any more details, that should be an ES, and if that is THAT ES, then these are fine score .

MM

the test above was meant to see whether the 25.6Ghz mem bandwidth for the 980 is actually limiting it,
the results shows a 2.5% difference in favor to the 1600Mhz vs the 1333.
that could be negligible though.
anyway,
show some Cinebench!

They used Cinebench 10( limited to 16 threads)
I used the new Cinebench 11.5 (24 threads) and scored 21.55
Screenshot in my thread in WCG section( It's here and OMG it's a beauty! )
Post #471..Good read.

[onex]

great post MM, especially the part with Loonym, very sad to hear the man story, hopefully Anabel is in good hands as said in his memorial thread, being taken cared of,
that's horrible to lose a father at such young age.

aside from that, and for the thread,

the case is awesome, seems a bit too big, maybe for dual MB system, yet you can't complain about having to send back that SM build even though it's still a pain u haven't got the chance to keep it ,

anyway, the bench is outstanding, 21.5 vs 11.58 on a dual W5590!
so insane it actually seems bogus .

[Movieman]

great post MM, especially the part with Loonym, very sad to hear the man story, hopefully Anabel is in good hands as said in his memorial thread, being taken cared of,
that's horrible to lose a father at such young age.

aside from that, and for the thread,

the case is awesome, seems a bit too big, maybe for dual MB system, yet you can't complain about having to send back that SM build even though it's still a pain u haven't got the chance to keep it ,

anyway, the bench is outstanding, 21.5 vs 11.58 on a dual W5590!
so insane it actually seems bogus .

You have to remember that the 11.58 on the W5590's was at 3458 and my 21.55 was at 4400..Big difference.

[onex]

the main difference is that the gainstown systems couldn't get much OC except from a single processor setup,
the ability to OC xeon's on a DP MB is really something to appreciate.

compare that to what has been few years ago with the SK system when every QX was priced 1750$ and had to be assembled on an entirely different 771 socket!
&150W power consp. not HT etc. etc.

things are much more open and better currently.

big difference .

[informal]

MM those 15K are 2xGulftowns @ 4.35Ghz,right? So estimated score @ 2.2Ghz(the presumed highest possible clock duploxx's opteron ran at- SE is 2.3Ghz) gives us roughly 15K/(4.35/2.2)=7.6K or roughly 10% better per clock and per core than Opteron.. I used 2.2Ghz for the 12c Opty as it is the best case scenario(for Gulftown's advantage).If it is clocked lower (sub 2.2Ghz),the 10% gap is gone. Pretty good showing from the latest Magny Cours .

PS feel free to correct me if my math is wrong