Intel Xeon 5600-Series: Can Your PC Use 24 Processors?

[falchard]

The thing I like about the G34 is Quad Channel Memory. I wonder why AMD would pass up the chance to show some competitive results against Intel unless they are planning to show something that will completely blow up the industry.

I don't see any fanboyism in that statement.

 

[kansur0]

I have written in the forums to ask that workstation and consumer graphics cards be tested to see if performance differs between cards. My request has not been answered.

Before this test 3D applications were the only applications that require drivers to work specifically with OpenGL or DirectX in order to display projects in realtime. Some 3D applications now have raytracing engines that will use graphics cards to render the final output. This still requires workstation cards and specific drivers such as CUDA to take advantage of the immense mathematical horsepower of the GPU.

Now that they are doing tests with Adobe CS5 it would be great to see if you really do need to buy a Quadro card to get maximum acceleration. Any Quadro card vs. the Fermi 480. The same could be done for ATI cards. FireGL vs. Radeon equivalent.

So? If anyone from Tom's Hardware is reading this for content other than just moderation...here is the request:

WORKSTATION VS. CONSUMER VIDEO CARD...ADOBE CS5 BENCHMARKS

If there really is a difference...explain it to us. What makes these two cards different. Break it down. Why can I use a consumer GPU and still get acceleration? Are workstation cards more efficient? Are consumer cards crippled? Where does OpenCL fit in the picture? (that could be a separate article)

When you press render in Adobe After Effects...what happens? The program sends the instructions to CPU and GPU. If GPU is present with this card and driver it renders faster because of a certain driver.

 

[Guest]

Why not review Intel's X7560 Xeon - it has 8 cores/16 threads

 

[hundredislandsboy]

Where's AMD's Opterons benchmarks or even Thuban's? I'm hoping you can prove that the difference is less than 10% while Intel's prices are 500% more than the 1055T.

 

[yodoso]

This is a server processor, where are server benchmarks?

 

[Guest]

is there any way to use these processors on x58 based motherboards as both of them have the same socket. And we won't be using the 2nd quickpath connector of the xeons, does it really matter that the 2nd QPI is also connected.
I mean a dual socket CPU on x58 based motherboard.

 

[cangelini]

[citation][nom]kansur0[/nom]I have written in the forums to ask that workstation and consumer graphics cards be tested to see if performance differs between cards. My request has not been answered.Before this test 3D applications were the only applications that require drivers to work specifically with OpenGL or DirectX in order to display projects in realtime. Some 3D applications now have raytracing engines that will use graphics cards to render the final output. This still requires workstation cards and specific drivers such as CUDA to take advantage of the immense mathematical horsepower of the GPU.Now that they are doing tests with Adobe CS5 it would be great to see if you really do need to buy a Quadro card to get maximum acceleration. Any Quadro card vs. the Fermi 480. The same could be done for ATI cards. FireGL vs. Radeon equivalent.So? If anyone from Tom's Hardware is reading this for content other than just moderation...here is the request:WORKSTATION VS. CONSUMER VIDEO CARD...ADOBE CS5 BENCHMARKSIf there really is a difference...explain it to us. What makes these two cards different. Break it down. Why can I use a consumer GPU and still get acceleration? Are workstation cards more efficient? Are consumer cards crippled? Where does OpenCL fit in the picture? (that could be a separate article)When you press render in Adobe After Effects...what happens? The program sends the instructions to CPU and GPU. If GPU is present with this card and driver it renders faster because of a certain driver.[/citation]

This is in the works! =)

 

[lauxenburg]

The eff? What about the new Opty chips, huh?

 

[hundredislandsboy]

I would like to suggest TH build a budget, midrange, and high end home video editing PC, not a fancy workstation costing $10,000 but something reasonable for the basic home video hobbyists, ie BDay parties, sporting events. And then compare dual core, quad and hexacore all the way tro Intel's $1,000 CPU.

 

[cangelini]

[citation][nom]lauxenburg[/nom]The eff? What about the new Opty chips, huh?[/citation]

Discussed on the first and last pages of the story AMD decided to sit this one out.

 

[shovenose]

nice review...
but dont you think that a 10ghz dualcore would outperform a 3ghz hexacore?

 

[cangelini]

[citation][nom]shovenose[/nom]nice review...but dont you think that a 10ghz dualcore would outperform a 3ghz hexacore?[/citation]

In iTunes, perhaps. In a threaded app, that's debatable.

 

[WarraWarra]

LMAO $1663 I can buy how many cpu's / cores + cluster them together for that price in a HPC.

I can sure as hell end up with more than 6/12 cores and 8+ times the performance for that price.

$1000 for AMD 6 cores = 5 $199 AMD Phenom II X6 1055T Thuban 2.8GHz + $663 for mboards + ram and use some parts I have already.

LOL 10x Intel Pentium E6700 $100.00 ect.

$1000 for Intel quad's = 6x q8300 $150 + $763 for other parts + stuff I already have.

I love it when they smoke some good stuff in California while surfing and coming up with this prices / products, just like the dream of 4MB/s broadband in the USA.

LOL this is not Africa with 8MB/s Adsl or avg. 25MB/s slow Adsl Europe.

 

[tu_illegalamigo]

Wonder how it runs Inventor or Solidworks, I like responsive, but wonder if that would be over-kill.

 

[tu_illegalamigo]

[citation][nom]WarraWarra[/nom]LMAO $1663 I can buy how many cpu's / cores + cluster them together for that price in a HPC.I can sure as hell end up with more than 6/12 cores and 8+ times the performance for that price.$1000 for AMD 6 cores = 5 $199 AMD Phenom II X6 1055T Thuban 2.8GHz + $663 for mboards + ram and use some parts I have already.LOL 10x Intel Pentium E6700 $100.00 ect.$1000 for Intel quad's = 6x q8300 $150 + $763 for other parts + stuff I already have.I love it when they smoke some good stuff in California while surfing and coming up with this prices / products, just like the dream of 4MB/s broadband in the USA.LOL this is not Africa with 8MB/s Adsl or avg. 25MB/s slow Adsl Europe.[/citation]

Dont think Beowulf clusters and the like are the same kind of workload. Plus the size/power considerations, running software that can take advantage of that kind of parallelization, some of these programs are single threaded.

 

[jecastej]

I have the same 8 core Xeon since 2008 (yes a Mac Pro). In 3 years (2006-2008) I went from one dual-core 2.3 PPC machine to an 8 core Xeon and it was a complete system for less than 3K, every time doubling performance. But today this i7 Core 870X with 6 cores and 12 treads is near the same 3K and it only offers a 30% increase in performance compared to the system I have (this 8 core does 6.67 in CINEBENCH11). This, in my opinion, shows that at the same price point Intel became "inefficient" or did not "scaled well" since 2008. Today I need to invest 5-6K, double the money to double my performance, yikes! AMD come to the rescue please!!... And this is why this year I am evaluating Octane CUDA GPU rendering software for up to 50X the performance from my current machine. $150-300 for the software and $500 - 600 for a Nvidia 480 game card. If I want to scale my system, 2 or more nvidia cards with different specifications could be added. I am sure I will lose manny Mental Ray rendering options but I can't miss this kind speed power. I can't ignore it for the price and features. If this works I may even "downgrade" on cores and buy a fast Quadcore machine with all the PCI express 16 lanes slots available. Crossing fingers on my Octane tests (see if this works for me).

In the end, I don't see the benefit to increase my investment 2 times today to have 2.5X the performance compared to this huge parallel GPU solutions at up to 50X gain. I may not see the entire 50X increase, but I will be happy with 10X and more.

 

[turok]

Although Lightwave 3D is a good program, there's greater chance that the more expensive 3D modeling and Animation programs would have better support for dual Xeons. Would have preferred to see you test the dual xeons with the heavyweights such as Maya 2011, Softimage XSI, 3DS Max 2011. Movie companies and most game companies use the heavyweight programs instead of Lightwave.

So this article didn't really help those interested in Workstations for 3D modeling and animation

 

[warmon6]

[citation][nom]wotan31[/nom]I'll bet Windows won't work with this many processors. The crap OS will probably BSOD. Even it boots, Windows is a virtual retard when it comes to thread management - it scales VERY poorly once you go above 4 cpu's. Linux or OSX on the other hand, would definitely benefit from such technology, since both of those have advanced thread management.[/citation]

[citation][nom]wotan31[/nom]Correct... if the software you use is Windows. Use a real OS that's based on UNIX and can actually scale properly when you give it serious hardware. Windows is a tinker toy in comparison.[/citation]

Dude, you just failed. Big time....

 

[Guest]

I think people are overlooking the fact that the purpose of the x5600 series is really designed for virtualization. you use 2 of these processors with HT on an esx box with a boatload of memory and you have yourself a datacenter in a box.

 

[Guest]

What you might do in a next test is setup some vm software, and run the software on 2 or 4 windows installations installed on the xeon system.

I would love to see what happens than.

 

[Guest]

> Hyper threading was kind of cool back in the P4 days

Hyper-threading today has very little in common with hyper-threading from the P4 days... it's vastly improved on several accounts.

 

[eddieroolz]

A great workstation review, as usual from Tom's Hardware.

 

[Dual1366Xeon]

How about showing some actual CPU usage in these benchmarks? I have a dual 1366 setup with (2)E5520's. I can run thread optimized software and have A LOT of other things running at the same time - with no performance hit. Win7 is very responsive no matter what I'm doing; hence the reason I built this...multitasking.

 

[Mr Pizza]

not REAL cores no thanks

 

[Guest]

I have two thoughts about this:

I would be very interested to compare these benchmarks with hyperthreading disabled. As good as Intel's hyperthreading gets, with heavy workloads it must be detremental to performance- the context switching causes the registers to be changed which is an expensive operation.

Secondly, these are not production processors yet, so I would imagine that optimizations will happen to the BIOS on the motherboard.

In conclusion, I was hoping for more in this review.