ASUS Z9 PE-D8 WS Dual Xeon Insanity E5-2660 & E5-2687W

very pleasing review mate
smile.png


this beast is superb..... with that price range, it's worthed
smile.png
 
A couple of weeks ago I was going to buy this board and two E5-2620's but then I read a review of the E5 chips inside an Asus Z9 PE-D8 WS and spoke to Jacob from EVGA's Forums. Both of those sources of information indicated the best overclock possible on the new XEON's was a lift in the BCLK from 100MHz to 105MHz maximum (and even that was not a sure thing). This would result in only a 100MHz improvement on the E5 2620's

It is quite disappointing that Intel has completely locked the new XEON processors. Some on the EVGA forums (who are pining over the SR-X) are planning to wait and see if Intel releases an X series XEON at a $1200+ price tag just for the benchmarking crowd but there is no indication Intel is even considering that and I decided not to wait for something I don't think will ever come to fruition.

In the end I picked up an Asus Rampage IV Extreme and 3930K which I've overclocked considerably. I don't regret that purchase one bit especially after hearing your conclusion in your review of the Asus Z9 where you basically say do what I just said if you want to overclock or build a killer gaming rig.
biggrin.png
 
If Rodney saw it he'd definately say it was a 100% Kick Ass product. Mind you he says that about everything LOL
tongue.png


Looks amazing by the way your gonna need your own sub station soon
 
imho the Z9PE-D8 WS can run 4 pcie at full 16x speed with a dual cpu setup.

afaik two xeons on this board provide 80 pcie lanes.

cpu 1:

slot 1 and 3 at 16x speed, slot 2 and 4 at 8x speed, slot 3 is downgraded to 8x if slot 2 and 4 are running 8x cards.

cpu 2:

slot 5 and 7 at 16x speed, slot 6 at 8x speed.

asus say on their website 2*16x / 4*8x, imho this is true with a single cpu setup.

with dual cpu this should be 4*16x, sadly i dont remember where i did read that...

it just sounds odd to have only 4*8x with 80 lanes.
 
[font=Arial, sans-serif]This was a lot of fun to digest. Thanks so much!. I know I love seeing the extreme boundaries, even tho they may not be so relevant to my own reality.[/font]

[font=Arial, sans-serif]I am not a gamer; but an artist/student of motion graphics & visual effects. Guys with the skills I'm learning build the games, the commercials, the movies, the broadcast graphics... Boring workstations heve been my life. My activity on a box mixes high resolution digital video, photos, 3D modeling, 3D animation, 3D motion tracking, and compositing. And my workflow demands hefty snap in real time tool response, previewing, and rendering. The proven gaming strategy of overclocking i7s/RAM, and stacking up multiple cuda GPUs would only get me so far.[/font]

[font=Arial, sans-serif]A solid WS strategy calls for a beefy balance of lotsa cores, lotsa RAM, SSD caching, RAIDed HDDs, a high-end GPU with "Adobe-blessed NVIDIA sensibility",and effective tweaking of graphics apps for performance. [/font]

[font=Arial, sans-serif]It's clear to most that a WS improves with the number of cores; but there is an interesting dilemma there. Graphics peeps fortunate enough to have a lot of cores (8+) can't often use them all. Oftentimes they encounter a phenomenon where applying too many cores actually slows their processing down (e,g, Adobe apps like After Effects). This happens when threading spawns many instances of a graphics program and cores begin to step on each other's toes as they compete for inadequate RAM. Installing more (and faster) RAM helps; but the biggest problem is that the multi-threading strategies the apps are coded with are inadequate or nonexistent. Benchmark apps are better in this regard, so they aren't as real world as we'd like. To their credit Adobe and others do enable their code to be assisted to some extent by OpenGL, Open CL, and Cuda processing in GPUs. But rendering is still not what it could be yet. So talk of processing cores and NVIDIA's Maximus schemes are really more tantalizing than satisfying to me. In spite of this, there still are reasons to invest in lotsa cores: 1) a typical graphics workflow requires many hefty apps to be open and processing simultaneously, 2) some rendering apps are threaded decently enough to capably use how ever many cores you enlist, and 3) multi-core machines built today will be able to fight obsolescence admirably as more and more apps jump on the multi-threading bandwagon.[/font]

[font=Arial, sans-serif]The system in the above test would problee make my life a whole lot easier!! But unfortunately I'm on a lower budget and looking to build a WS. Here's what I'd really love-LOVE to see at Overclock3D (and I'm sure you've thought of it already...) I'd love to see a more practical shootout based on the cost/performance expectation of the more typical graphics grunt. Maybe start with a baseline single CPU config based on the E5-2687w tested here, and then compare it to the same amount of cores achieved by a dual E5 Xeon config that costs roughly the same. And for good measure throw in a config that's cheaper but has even more cores....[/font]

[font=Arial, sans-serif]Certainly we're still talking more cost here than the high-end gaming MoBo/CPU, but less cooling. And how awesome would it be to experience the effectiveness of Intel QPI, plus chew on the subtle differences in CPU cache and clock-speed! Here would be my contenders:[/font]

[font=Arial, sans-serif]------------------------------------[/font]

[font=Arial, sans-serif]1) Single CPU, 8-core (baseline)[/font]

[font=Arial, sans-serif]Single E5-2687w Xeon CPU ([/font][font=Arial, sans-serif]8 core, 3.1 – 3.8 GHz, 150w, 20MB cache)[/font]

[font=Arial, sans-serif]ASUS P9X79 WS motherboard[/font]

[font=Arial, sans-serif]Cost: $2280 US[/font]

[font=Arial, sans-serif]------------------------------------[/font]

[font=Arial, sans-serif]2) Dual CPU, 8-core[/font]

[font=Arial, sans-serif]Two E5-2643 Xeon CPUs [/font][font=Arial, sans-serif](4 core, 3.3 – 3.5 GHz, 130w, 2 QPI Links, 8GT/s, 10MB cache)[/font]

[font=Arial, sans-serif]ASUS Z9PE-D8 motherboard[/font]

[font=Arial, sans-serif]Cost: $2370 US ($90 more than the baseline strategy)[/font]

[font=Arial, sans-serif]------------------------------------[/font]

[font=Arial, sans-serif]3) Dual CPU, 12-core[/font]

[font=Arial, sans-serif]Two E5-2630 Xeon CPUs ([/font][font=Arial, sans-serif]6 core, 2.3 – 2.8 GHz, 95w, 2 QPI Links, 7.2GT/s, 15MB cache)[/font]

[font=Arial, sans-serif]ASUS Z9PE-D8 motherboard[/font]

[font=Arial, sans-serif]Cost: $1820 US ($460 less than the baseline strategy)[/font]

[font=Arial, sans-serif]------------------------------------[/font]

[font=Arial, sans-serif]AGAIN, Thanks for the very intriguing & enlightening test. Very helpful for us WS peeps![/font]

[font=Arial, sans-serif]Dayoldy[/font]
 
One E5-2687W in Cinebench

I'm about to buil a rig for videoediting in Adobe cs6 so I guess that the cinebench is the benchmark for me? Due to the expensive cpu-cost in a dual E5-2687W setup I have an idea to cut the initial cost by beginning with one cpu and keep the option to upscale to two cpus if needed in the future.

The question is how will on E5-2687W stack up to an i7-3960X setup in Cinebench?
 
imho the Z9PE-D8 WS can run 4 pcie at full 16x speed with a dual cpu setup.

afaik two xeons on this board provide 80 pcie lanes.

cpu 1:

slot 1 and 3 at 16x speed, slot 2 and 4 at 8x speed, slot 3 is downgraded to 8x if slot 2 and 4 are running 8x cards.

cpu 2:

slot 5 and 7 at 16x speed, slot 6 at 8x speed.

asus say on their website 2*16x / 4*8x, imho this is true with a single cpu setup.

with dual cpu this should be 4*16x, sadly i dont remember where i did read that...

it just sounds odd to have only 4*8x with 80 lanes.


You did read it on there website (http://uk.asus.com/Motherboards/Intel_Socket_2011/Z9PED8_WS/#specifications) under specs. it does say you can run 4 PCI-E lanes at the full 16x.

"4 x PCIe 3.0/2.0 x16 (dual x16 or quad x8)
2 x PCIe 3.0/2.0 x16"

so according to the first line 2 16x can be run and at the same time 2 more 16x lanes can be run. Its only after that that speed starts to go down.

it will be good to see how the up and coming quadro cards will work with this...
 
"14 SATA ports (8 SATA 3Gbp/s and 6 SATA 3Gbp/s)"

Little typo :D It should be: 14 SATA ports (8 SATA 3Gbp/s and 6 SATA 6Gbp/s)
 
Hi Tom,

Just read this review. Awesome. Looking at a Xeon system, but seriously, the cost. /cry

You were commenting on getting 3d software to show the difference in performance.

Well, http://www.blender.org/download/get-blender/ is a link for open source 3d software.

The test scene used by the community: dl.dropbox.com/u/1742071/1m/BMW1M-MikePan.blend the scene is ready to render, just hit f12.

Original link on http://blenderartists.org/forum/sho...Cycles-render-benchmark&highlight=cycles+test

I would expect you doing a mini review showing the differences in the speed of various setups would generate a lot of interest from the blender community. :)
 
I love the review.

I remember that you did a review of the Xigmatek Elysium.

Would this board fit w/2 corsair h80s at the top?

The board is EEB format. The Elysium holds E-ATX. Both formats are 12" x 13". Are the standoffs in different positions?

Thanks!
 
Back
Top