• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Dual Xeon 5080's will blow Conroe Away

meansizzler said:
Anyone seen the specs on these, a Single Xeon 5080 which is a Dual Core 3.73GHZ CPU with Hyper Threading and 1066MHZ FSB , even one of these chips is enough to blow away an X6800, and two of them will smoke it...wonder if anyone on these boards have one...

http://www.intel.com/products/processor_number/chart/xeon.htm
Click to expand...


lol those are dempsey processors and their crap, i dont no why intel releasied them because no 1 is going to buy them when a woodcrest came out only a few weeks later.

dempsey = 2 prescott's
woodcrest = 2 conroes
 
The woodcrest in based on the Conroe Architecture and at 3 GHZ and 1333MHZ FSB and the Dual Core it would own every CPU out there, and thats just one of these...and considerign people are paying £600 for an X6800 a £1000 does not seem much anymore..., well not if your into video production...
 
Hmm.. Xeon 5140 gives the same performance if not better than the Xeon 5080 and its only £500, far cheaper than a X6800, Xeon 5080 retails at £820...so really not that much...
 
The woodcrest have a 1333mhz fsb and 2 independent buses. You can really feel the difference in games that use a lot of threads like Oblivion. With 2 cores (or quad core) Windows (at least Windows 64 or Windows 2003 Server) feels a lot more responsive.

The fb-dimms are good if you are doing a lot of number crunching as they provide a hell of a lot more bandwidth then ordinary ddr2 which is good for what I do, but the downside is high latency due to the translation between the fb chip and ddr2 . They are not really designed for gaming. However the extra bandwidth more then makes up for the fact that in certain benchmarks like 3dmark06 I am scoring about 10,700. With ordinary ddr2 at tighter timings it would be closer to 11k. But benchmakrs don't matter to me so much.


BillytheImpaler said:
Has anybody run a Conroe at a 1333 MHz FSB to see what diffference that fancy new memory makes for the Woodcrests?
Click to expand...
 
One website has looked at the difference between the top-end Woodcrest Xeon and the X6800 core duo and the result - given the higher FSB and operating frequency of the Xeon - are quite surprising. In pretty much all the gaming benchmarks, the Core Duo Xtreme wins by an appreciable margin and it goes on to dominate in a lot of the workstation-class ones, too.

The reviewers attribute this discrepancy with the available specs to be down to the huge latency that the FB-DIMMs introduce in comparison to the already lethargic DDR2. So while the Xeon chips may be faster on paper and somewhat cheaper than the top spec Core Duo, the higher cost of the motherboards and RAM to stick in them, coupled with the lower performance, makes them a bit of a non-starter for home users.
 
I don't believe the Core 2 Duo extreme is faster. At least not on the benchmarks my mates have run. It just does not make sense a cpu with a 1333fsb could be slower. It is definitely the fb-dimms as you say, but the difference I think would be marginal in real world terms e.g gaming. The extra cpu and sheer bandwidth of the fb-dimms more then makes up for the higher latency. The trick is to get the right mix of fb-dimms - too many and latency suffers as the more you add the higher the latency since memory requests start on stick1 and work down from there. And too little and you reduce bandwidth and not run in quad mode at all. The sweetspot is 4 sticks of FB-DDR2.

My mobo will take regaular ecc ddr2 which I am borrowing soon to test with.
 
Right, here's a rundown of the results. Apologies to the website for reprinting the gist of its testing without permission. ;)

Sandra 2007: The Conroe X6800 beats the Xeon 5160 in the CPU stakes, but gets beaten in the Multimedia test. Neither win is what I would describe as significant. The Conroe wallops the Xeon platform in terms of memory bandwidth - almost twice as much - as well as 14% less latency. Chalk this one partially up to the DDR-800 for the Conroe as opposed to the DDR-667 on the Xeon...

Half Life 2, Ep.1: The Conroe X6800 take a 17.5% lead over the Xeon...

Fear: an enormous 1.7% lead for the Conroe...

Maya/3DSMax: The Xeon takes a 1s and 3s lead, respectively...

Lame 3.97: Both are equally brutal when encoding sound files...

WMP9 encoder: The Conroe hits what looks to be a hard disk limitation with a awesome 6s win over the Xeon...

Adobe Photoshop CS filter testing: 3s faster on the Conroe

Macromedia Flash MPEG import: Conroe is faster by 7s...

7-Zip file compression: even the E6400 is faster than the Xeon 5160... by nearly 14%.

Winrar: Xeon is quicker this time, but Conroe X6800 is still 23% faster...

Superpi and Cinebench are exactly equal, however.




In the tests run, the Xeon certainly is not slow by any stretch of the imagination. But it is slower than the Conroe in almost all the tests. And it will be more expensive from the initial outlay, too. Apache, however, might show a completely different story as well as other programs. But when it comes down to games, Conroe is still king. By a long way.

I would post the test, but I'm not sure if it is classed as a competitor, so better safe than sorry. ;)

Apologies if this post comes up as pendantic and childish, but it's late and my sense of humour isn't working at this point in time. Normal service will be resumed tomorrow. :D
 
Interesting. Is that from Ananadtech ?

Will be benching with ecc ddr2 soon to see if this makes a difference over fb-ddr2 but I really doubt it as I am running with 4 sticks in quad mode and the memory bandwidth being reported in sandra is well over 6k which is what my mate gets with his Conroe setup. Maybe the latency and possibly immature chipset drivers are the only thing that could explain the difference in the gaming benchmarks. Either that or the way they configured their setup. I still find it hard to believe a chip with a 1333fsb and dual independent buses could be slower then a 1066fsb chip. Regardless running 2 of these Woodcrests certainly picks up any slack. Only subjective of course, but games like Fear certainly feels faster

Also 667mhz v 800mhz memory does not come into it. The top of the range Conroe chip with it's 1066mhz fsb does 8.5GB/s, the Woodcrest a little more. DDR2 6400 memory that runs at 800mhz has a peak of 12.8GB/s. The Intel 5000x chipset is limited to DDR2 5300 667mhz though, which is 10.6GB/s but still more then enough for Conroe/Woodcrest is capable of at least at stock speeds.
 
Last edited:
No, it's not Anandtech - it's an online OEM that also have a 'lab' section. I find their reviews very objective, generally, as they just review the stuff the passes through their shop. As such, I'm not going to post the site here.

The thing with the FB-DIMMs is that - due to the additional step in writing/reading the memory - the latency of the modules is significantly higher. Now, as they are meant to be capable of much higher speeds (in future) the latency will cease to be such an issue. The only hiccup in such a plan is that it's taken DDR2 almost an effective doubling of MHz-age to account for its slacker memory timings over DDR400 on tight timings. All the bandwidth in the world is of little use if it takes the CPU/memory controller forever to do anything with it.

The system might appear smoother as a result of the two-sided FSB, for the smae reason that using SMP is always smoother than single chip solutions. I don't know as I've never used either dual core or dual-FSB: hell, I've never even used an Athlon 64 system...
 
If the ecc ddr2 I am testing next week works out to be faster then I may go that route.

<edit>Must have been smoking something....it looks like my mobo will not handle ordinary ecc ddr2. So will have to put up with latency issue. From what I have read on Anand 4 fb-dimms is the optimal as it gives good bandwidth in quad mode and decent latency. And more then 4 and latency takes an even bigger hit</edit>
 
Last edited:
Flanno said:
If the ecc ddr2 I am testing next week works out to be faster then I may go that route.

<edit>Must have been smoking something....it looks like my mobo will not handle ordinary ecc ddr2. So will have to put up with latency issue. From what I have read on Anand 4 fb-dimms is the optimal as it gives good bandwidth in quad mode and decent latency. And more then 4 and latency takes an even bigger hit</edit>
Click to expand...

lol i thought that sounded wrong, but i didnt want to argue as you have / know the board :p (i thought)

give a try of these benchmarks, would be good to see how it compares to my processor / others.

http://forums.overclockers.co.uk/showthread.php?t=17628286

http://forums.overclockers.co.uk/showthread.php?t=17629198
 
On Fritz's Chess Benchmark I get
19.89 / 9254 - everything at stock.
If was able to overclock the cpu's I am sure I would get another couple of thousand :)

<edit>About 16k on the nuclearMC score...was running bitcommet and doing a virusscan at the time :) </edit>
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom