Xeon W3520 3845A935 - DFI UT X58-T3EH8

eva2000

Gawd
Joined
Apr 14, 2002
Messages
539
My first W3520 and nice week batch 3845A935

w3520_3845a935_tn.jpg


4000Mhz at 1.184v load = coretemp = 57-60c load temps in 26.5C ambients!

prime95_customlarge_load.png


4446Mhz at 1.360v
Very strong on BCLK'ing even on 20x multiplier!

spi32m_8m17s079ms.png


4534Mhz at 1.376v

spi32m_8m07s470ms.png


4600Mhz at 1.472v

Bumped up vcore and tried 20x230bclk = 4600Mhz and 32M Pi passed easily :D Haven't tried lower vcore yet though :)

spi32m_8m02s602ms.png


  • Intel Xeon W3520 3845A935 D0 (Xeon equivalent of Core i7 920 D0)
  • EK Supreme LT 1366 / FESER 480 / MCP655
  • DFI UT X58-T3EH8 04/28 beta
  • 2x HD4870x2 Crossfire (Powercolor + Gigabyte flashed to Asus Top bios) with ATI Win7 9.4 (8.612 WHQL)
  • 6GB Corsair Dominator GT DDR3-2000Mhz 7-8-7-20
  • 750GB Samsung SATAII
  • Silverstone OP1200 PSU
  • Triple boot - WinXP Pro SP3 / Vista Ultimate 64Bit SP2 / Win7 Ultimate 64bit RC









Time to really push my W3520 on H20 cooling still. Managed to break alot of personal best benchmarks for max validation, max uncore frequency achived, Super Pi 1M, 32M and Pi Fast on i7/X58 platform :)

  • Super Pi 32M HT Enabled max = 4677Mhz at 1.488v
  • Super Pi 32M HT Disabled max = 4794Mhz at 1.552v
  • CPUZ Validation HT Disabled max = 4877Mhz at 1.584v

System:
  • Intel Xeon W3520 3845A935 D0 (Xeon equivalent of Core i7 920 D0)
  • EK Supreme LT 1366 / FESER 480 / MCP655
  • DFI UT X58-T3EH8 04/28 beta
  • HD4870x2 (HIS 4870X2 to Asus Top bios) with ATI WinXP 9.5
  • 6GB Corsair Dominator GT DDR3-2000Mhz 7-8-7-20
  • 750GB Samsung SATAII
  • Silverstone OP1200 PSU
  • Triple boot - WinXP Pro SP3 / Vista Ultimate 64Bit SP2 / Win7 Ultimate 64bit RC

Max Validation @4877Mhz at 1.584v

Max Uncore Frequency @4645.6Mhz





Super Pi 1M & PiFast @4802Mhz at 1.584v

spi1m_8s406ms_tn.png


pifast_1753ms_tn.png


Super Pi 32M @4794Mhz at 1.552v

spi32m_7ms24s422ms_tn.png


Super Pi 32M @4677Mhz at 1.488v

spi32m_7m41s796ms_tn.png
 
Last edited:
air or water?

his sig says hes using a TRUE. if so thats pretty impressive. i heard of an i7 920 or 940 doing 4.8 on this dfi x58 board too. dfi has been shitty lately in terms of reliability and customer support, but man when their stuff works it WORKS.
 
I've got the same DFI board and D0 i920 Batch 3846B109, chip came back to the US with me from a business trip in Asia, apparently a good batch, but Newegg sent me a bad ATX case. To make a long story short I have everything running at stock or 3.2GHz on Passive cooling on a Thor's Hammer. (Thermal cycling the AS5). Hopefully I'll be able to put my numbers up next week. I'm dying in agony to see what this chip can do.

Great OC!!! Kudos OP!!!

All I'm looking for is a stable OC on air at 4.0-4.2GHz. hopefully I get some love from my D0.
 
I've got the same DFI board and D0 i920 Batch 3846B109, chip came back to the US with me from a business trip in Asia, apparently a good batch, but Newegg sent me a bad ATX case. To make a long story short I have everything running at stock or 3.2GHz on Passive cooling on a Thor's Hammer. (Thermal cycling the AS5). Hopefully I'll be able to put my numbers up next week. I'm dying in agony to see what this chip can do.

Great OC!!! Kudos OP!!!

All I'm looking for is a stable OC on air at 4.0-4.2GHz. hopefully I get some love from my D0.

stock is 2.66... nice OC for passive cooling. you gonna just put a 120mm fan on the thors hammer when you take it to 4?
 
Very nice clocks eva! 4GHz at 1.18 is impressive...very impressive. I just searched and found that PAo_ReVolt is selling that same cpu on ebay. So...guess what I gonna do?;)
 
you should PM him. hes a trusted seller here on the forums and you might be able to cnvince him to cut you a [H]ard deal. :)
 
@ Ekuest

Yeah I'm running stock (2.66) or 3.2 on passive, to cycle my AS5, getting it hot and then letting it run warm. My new case and the case this setup resides in currently have a different air flow pattern didn't want to slap on a fan for nothing, since I know I have to remove the mobo from case sometime soon. Newegg sent me a bad atx case, so I'm dying to OC this machine, but can't do it in an ATX mini tower. Not enough airflow.

@3.2GHz it's not killing my X3350 at 3.6GHz, but it's quicker in somethings.

Fans didn't come in yet. Push pull 2 75CFM fans for me, got a Swiftech WC setup lying around somewhere not being used, if the Thor can't take the heat. But a new waterblock is going to cost me the same as a good aircooler.

Wanted to go crazy with 100cfm scythes but no ball bearings on those.
 
Last edited:
Nice work Eva2000, that chip is exactly the same batch as mine. Here's a 24 hr prime95 test

and it's highest HWbot validation
yikes huge screenie.. resize would be nice hehe.. but yes this batch is crazy

EK Supreme LT /FESER 480 H20

3dmark06 @21x215bclk - 4509Mhz

3dmark06_830965_31403_turboi7_tn.png


3dmark05 @21x215bclk - 4509Mhz

3dmark05_852_1000_39765_tn.png


3dmark Vantage @21x215bclk - 4509Mhz

3dvantage_847_1000_28077_tn.png


i've got a 3901A250 will be testing it next week :)
WOW first ever 2009 batch! let us know how that goes :)

I've got the same DFI board and D0 i920 Batch 3846B109, chip came back to the US with me from a business trip in Asia, apparently a good batch, but Newegg sent me a bad ATX case. To make a long story short I have everything running at stock or 3.2GHz on Passive cooling on a Thor's Hammer. (Thermal cycling the AS5). Hopefully I'll be able to put my numbers up next week. I'm dying in agony to see what this chip can do.

Great OC!!! Kudos OP!!!

All I'm looking for is a stable OC on air at 4.0-4.2GHz. hopefully I get some love from my D0.
thanks... 3845B/3846B are great batches indeed!

Very nice clocks eva! 4GHz at 1.18 is impressive...very impressive. I just searched and found that PAo_ReVolt is selling that same cpu on ebay. So...guess what I gonna do?;)
grab them quick - buy a few if you can LOL

W3540 in 3845B010/017 are also gems apparently!
 
Damn; I wish Pao_Revolt had these last week when I ordered mine for $20 more. He has the 3845A935's; same as the one I got.

waiting on mobo to test.
 
Damn; I wish Pao_Revolt had these last week when I ordered mine for $20 more. He has the 3845A935's; same as the one I got.

waiting on mobo to test.

From one of his auctions: "The C0 Stepping has known issues and the new D0 Stepping has eliminated those."

What is he talking about?
 
Impressive vcor to say the least ! Though from your posts over the years I don't recall you doing much of it, but is that vcor prime stable ?
 
Doubt it's prime stable but it's wprime 1024m stable so pretty good :D

Clocked GPUs a bit higher at 862/2000 and equaled Dino's #3 placed score of 28,559 :D

3dvantage_862_1000_28559_14v_tn.png


3dmark2003 @21x215bclk - 4509Mhz

140k broken :)

3dmark2003_862_1000_140486_tn.png
 
Congrats on the PB's, just sitting at my desktop I started backepeddling the voltages, and it crashed at around 1.25v, 4503, lol
 
Are there other boards putting up monster o/c's with the Xeon other than Classified and DFI?
 
I'm pretty sure the "C0 issues" are really just "I have something that I can claim is newer and better, so I'm going to claim something is wrong with the ones that are easy to get for the sole purpose of extorting more money out of people" eBay seller issues.
 
Are there other boards putting up monster o/c's with the Xeon other than Classified and DFI?

Well there are your normal crop of X58 boards which normally hit the wall right at about 200MHz BCLK and some of those have individual samples hitting 205MHz BCLK or so.Then there is another tier of overclock friendly X58 boards, the EVGA X58 3X SLI Classified, the Foxconn Blood Rage and the DFI UT X58-T3EH8. I can't speak to the DFI board in question as I've never worked with them. One thing I've noticed is that the boards from ASUS, MSI, and Gigabyte don't respond well or at all to most of the settings in the BIOS. You can tweak many of the options all day and they will have little to no impact on your overclock. On the EVGA X58 3X SLI Classified I had to tweak more settings than I've ever had to adjust in order to get the overclock I achieved. I hit a 190MHz BCLK wall until I started tweaking more of the settings EVGA recommended adjusting in their own forum threads.
 
I emailed GigaByte about running a Intel Xeon W3520 on a EX58 board, they refused to talk about the cpu working correctly, but I see the top end Asus WS and MSI workstation boards run the CPU with no issue, and now a OC champ desktop DFI board
-
HardOCP for the win!
 
Posted on the front page and submitted to DIGG for a killer Xeon OC. :)
thanks just needs a correction that is 4.6ghz water cooled not air cooled ;)

The C0's have no issues that I am aware of.
Yeah, i think the poster was referring to the errata list posted by some folks for i7 c0 i.e. S1 state not resuming on some C0 which is fixed in C1. But board bios might play apart. I had no problems resuming from sleep state for vista 64bit for my 2x i7 920 c0 and 2x i7 920 d0. Haven't tried for w3520 but it's d0 anyway.


Haha, nice to see you on another forum I populate Eva :)
hehe, i get around :D

Nice!

so now it's gonna fold for 33 right? :cool:

Congrats on a nice OC
nah no folding too busy benching/oc'ing :D

Nice.

What kind of numbers are you pulling with the X264 benchmark?
haven't tried it since my first i7 920 c0 cpus.

Are there other boards putting up monster o/c's with the Xeon other than Classified and DFI?
Asus P6T WS Pro with certain bioses is closing in, in terms of max bclk that is.. only EVGA Classified, DFI UT X58-T3EH8 and Asus P6T WS Pro have seen >250bclk when sub zero cooled as far as i know.

Well there are your normal crop of X58 boards which normally hit the wall right at about 200MHz BCLK and some of those have individual samples hitting 205MHz BCLK or so.Then there is another tier of overclock friendly X58 boards, the EVGA X58 3X SLI Classified, the Foxconn Blood Rage and the DFI UT X58-T3EH8. I can't speak to the DFI board in question as I've never worked with them. One thing I've noticed is that the boards from ASUS, MSI, and Gigabyte don't respond well or at all to most of the settings in the BIOS. You can tweak many of the options all day and they will have little to no impact on your overclock. On the EVGA X58 3X SLI Classified I had to tweak more settings than I've ever had to adjust in order to get the overclock I achieved. I hit a 190MHz BCLK wall until I started tweaking more of the settings EVGA recommended adjusting in their own forum threads.
FYI, all 5 of my cpus hit 230bclk for 32m pi/3d benching on DFI UT X58-T3EH8 on air and water with validations ranging from 233-235bclk.

  • i7 920 3836A756 C0 CPU #1
  • i7 920 3836A756 C0 CPU #2
  • i7 920 3851A368 D0 CPU #1
  • i7 920 3851A368 D0 CPU #2
  • Xeon W3520 3845A935 D0
never used subzero on i7/x58 setup before, but seen with DFI UT X58-T3EH8 someone push 254bclk with i7 920 d0 IIRC http://valid.canardpc.com/show_oc.php?id=572978



I emailed GigaByte about running a Intel Xeon W3520 on a EX58 board, they refused to talk about the cpu working correctly, but I see the top end Asus WS and MSI workstation boards run the CPU with no issue, and now a OC champ desktop DFI board
I've seen a few W3520/W3540 results on Gigabyte X58 boards so they do work, although limited by bclk or around 222bclk unless you use 128-130mhz PCI-E frequency with special Gigabyte X58 bioses which aren't available to the public yet (testing stage).
 
I emailed GigaByte about running a Intel Xeon W3520 on a EX58 board, they refused to talk about the cpu working correctly, but I see the top end Asus WS and MSI workstation boards run the CPU with no issue, and now a OC champ desktop DFI board
-
HardOCP for the win!

Whoever responded to you, didn't know their product very well ! Here are some screens when I first received the W3520. I was able to push the UD5 up to 221 for suicide runs, but 210-215 was perfectly usable from the 4 series beta bios forward. All options were unlocked just like any 920
http://www.xtremesystems.org/forums/showpost.php?p=3750261&postcount=339

edit: actualy if you explore that thread, you will see that Asus users were having a lot of probs initialy and thought it was chip related, but were happy to see that Gigabyte supported it early on, meaning just a bios rev. After first getting the Classifed, I was limited to 2:6, and 2:8 on the shipping bios, when mentioned in their forums they had it corrected within a day or so.
 
I've seen a few W3520/W3540 results on Gigabyte X58 boards so they do work, although limited by bclk or around 222bclk unless you use 128-130mhz PCI-E frequency with special Gigabyte X58 bioses which aren't available to the public yet (testing stage).

Is that part of what Dinos22 was mentioning he was working on with Gigabyte ?
 
I can't speak to the DFI board in question as I've never worked with them.=

I've been using DFI for four builds Opteron 170, E8400, Xeon X3350, and now my i920. Haven't had results good enought to post with on the 920 yet, because I'm still waiting on the case that I'm going to put this rig in, but I can say that the boards DO OC well.

I don't believe that they have a US based support team anymore and you have to go to a forum like "DFI Club" or something like that for any support.

They're not for everyone and not as feature rich as the Gigabyte boards, but if you want a simple board that's focused on OC'ing and you've got a good PSU, they tend to work well. Just my personal experience.

Also I wanna give props to EVA2000, he's posted all this info on the DFI board, he wrote up a guide on i4memory.com and it's really helpful for OC'ing with it. [H] forum is still my primary, cuz funny shit get said here.

Back on topic. Even the DK series of DFIs (which is using analog PWMs) OC's pretty well. I haven't had a board with digital PWM until the DFI UT X58-T3H8. Hope it makes a difference.

The only update I have on my 920 was that I got it to post at 4.0GHz at stock volts, last night, but no boot. Still passive on the Heatsink, so don't wanna fry it.
 
Last edited:
I'm thinking my W3520 could hit similar results... unfortunately my current one is in the DFI LP JR X58 in a SG03, so I'm stuck with stock cooling :(

I can hit 3.0 at 1.0v no problem, and I've done 4.0 on auto volts for a nice 9s SuperPi run, but I obviously can't dissipate enough heat to stay at 4.0.

I'm planning on going with a similar setup in my main system (sig) once I upgrade to i7:

W3520 / DFI UT X58 / 6GB OCZ DDR3-1600
 
Asus P6T WS Pro with certain bioses is closing in, in terms of max bclk that is.. only EVGA Classified, DFI UT X58-T3EH8 and Asus P6T WS Pro have seen >250bclk when sub zero cooled as far as i know.

FYI, all 5 of my cpus hit 230bclk for 32m pi/3d benching on DFI UT X58-T3EH8 on air and water with validations ranging from 233-235bclk.

  • i7 920 3836A756 C0 CPU #1
  • i7 920 3836A756 C0 CPU #2
  • i7 920 3851A368 D0 CPU #1
  • i7 920 3851A368 D0 CPU #2
  • Xeon W3520 3845A935 D0
never used subzero on i7/x58 setup before, but seen with DFI UT X58-T3EH8 someone push 254bclk with i7 920 d0 IIRC http://valid.canardpc.com/show_oc.php?id=572978

Thanks for the info and posting on the web your bios settings(?). Quickly looked over the info you posted on i4memory, pretty detail, wish all would do the same. Helps out the community.
 
I've been using DFI for four builds Opteron 170, E8400, Xeon X3350, and now my i920. Haven't had results good enought to post with on the 920 yet, because I'm still waiting on the case that I'm going to put this rig in, but I can say that the boards DO OC well.

I don't believe that they have a US based support team anymore and you have to go to a forum like "DFI Club" or something like that for any support.

They're not for everyone and not as feature rich as the Gigabyte boards, but if you want a simple board that's focused on OC'ing and you've got a good PSU, they tend to work well. Just my personal experience.

Also I wanna give props to EVA2000, he's posted all this info on the DFI board, he wrote up a guide on i4memory.com and it's really helpful for OC'ing with it. [H] forum is still my primary, cuz funny shit get said here.

Back on topic. Even the DK series of DFIs (which is using analog PWMs) OC's pretty well. I haven't had a board with digital POWM until the DFI UT X58-T3H8. Hope it makes a difference.

The only update I have on my 920 was that I got it to post at 4.0GHz at stock volts, last night, but no boot. Still passive on the Heatsink, so don't wanna fry it.

I've worked with plenty of DFI boards dating back to the mid to late 1990's. I'm not a fan of them as a general rule. I can't speak to the board in question though as I've never used one. We don't often review DFI boards on the [H].
 
Pretty please, with sugar on top, can you run x264 and let us know what your results are? :)
may do it but will be much later hehe

Really impressive OC, and good voltage.
thanks :)

Is that part of what Dinos22 was mentioning he was working on with Gigabyte ?
Yeah what Dino was mentioning - he's in Taiwan now for computex and GOOC 2009 finals so probably not testing much now.

I've been using DFI for four builds Opteron 170, E8400, Xeon X3350, and now my i920. Haven't had results good enought to post with on the 920 yet, because I'm still waiting on the case that I'm going to put this rig in, but I can say that the boards DO OC well.

I don't believe that they have a US based support team anymore and you have to go to a forum like "DFI Club" or something like that for any support.

They're not for everyone and not as feature rich as the Gigabyte boards, but if you want a simple board that's focused on OC'ing and you've got a good PSU, they tend to work well. Just my personal experience.

Also I wanna give props to EVA2000, he's posted all this info on the DFI board, he wrote up a guide on i4memory.com and it's really helpful for OC'ing with it. [H] forum is still my primary, cuz funny shit get said here.

Back on topic. Even the DK series of DFIs (which is using analog PWMs) OC's pretty well. I haven't had a board with digital POWM until the DFI UT X58-T3H8. Hope it makes a difference.

The only update I have on my 920 was that I got it to post at 4.0GHz at stock volts, last night, but no boot. Still passive on the Heatsink, so don't wanna fry it.
thanks for the kind words :) Oh who needs pc cases ? Haven't used one in 5+ yrs besides my HTPC rig hehe. Too hot in Brisbane here - room temps in summer can get as high as 33-38C here. Luckily it's cold season so 20-26C nights and 26-30C days right now :)

I'm thinking my W3520 could hit similar results... unfortunately my current one is in the DFI LP JR X58 in a SG03, so I'm stuck with stock cooling :(

I can hit 3.0 at 1.0v no problem, and I've done 4.0 on auto volts for a nice 9s SuperPi run, but I obviously can't dissipate enough heat to stay at 4.0.

I'm planning on going with a similar setup in my main system (sig) once I upgrade to i7:

W3520 / DFI UT X58 / 6GB OCZ DDR3-1600
yeah good cooling is a must :) what batch is your W3520 ?

Thanks for the info and posting on the web your bios settings(?). Quickly looked over the info you posted on i4memory, pretty detail, wish all would do the same. Helps out the community.
Glad I can help.. lots of fun testing stuff especially on memory side of things :D I'll be doing similar testing/guides etc for DFI JR X58, Gigabyte X58-UD4P and Asus PT6 WS Pro (fingers crossed on that one).

I've worked with plenty of DFI boards dating back to the mid to late 1990's. I'm not a fan of them as a general rule. I can't speak to the board in question though as I've never used one. We don't often review DFI boards on the [H].
Shame though... same here used DFI NF4 s939 series, 875P-T, DFI P35/X38/X48/X58 but skipped P45.


Some High Uncore frequency testing http://i4memory.com/f83/i7-x58-mem-freq-vs-mem-timings-vs-uncore-frequency-18156/ :D

everest_cputweaker_616265.png


spi32m_8m44s797ms.png
 
We don't often review DFI boards on the [H].

It's already been posted why Hard doesn't review certain bands, but on the side lines does Hard do any high to extreme oc's or use boards they intend not to review. There have been a few oc articles posted in the past but I didn't know what the norm is.
 
It's already been posted why Hard doesn't review certain bands, but on the side lines does Hard do any high to extreme oc's or use boards they intend not to review. There have been a few articles posted in the past but I didn't know what the norm is.

Honestly, we got over it a few years ago. We held the P4 OC world record for two years before it got popped. After that it was getting all commercialized, AMD and Intel hiring teams to tout their products with tens of thousands of dollars of support. Just was not fun anymore for us.

That all said, we are reviewing the EVGA Classified, and given it is truly built for extreme OCing, I felt it would not be right not cover some of the low temp features on the board, so we will be doing some dry ice OCing on it.

Still though, we were always more real world oriented OC site instead of I ran my CPU long enough to get an extreme screen shot kinda site.
 
Waves at Kyle :D

EVGA Classified without sub zero testing would be a crime as that's what the board is built for. Just remember to insulate the dimm slots as a few folks have killed quite a quantity of memory on EVGA Classified under extreme sub zero cooling :)
 
Back
Top