DFI Lanparty 875P-T LGA775 Benchmark Adventures

I was lucky enough to grab a P4 3.4E 550 cpu on Christmas Eve to go with my 2nd hand DFI Lanparty 875P-T motherboard that I bought off Oppainter at xtremesystems.org forums :) Had a few initial problems in that the system didn't boot up but loosening the G4 Storm waterblock's mounting screws a bit on the DFI Lanparty 875P-T allowed it to boot.

System:

P4 3.4E 550
SL7J8
Q433A423
Nov 26, 2004
Malaysia

DFI Lanparty 875P-T with 11/30 bios
2x 256MB KHX3200K2/512 BH-5
64MB Albatron MX480EL (for testing)
120GB WD 1200JB
Pioneer AO5 DVD Burner
580W Topower 686P6 Adjustable rail PSU

CPU cooled by:

Cathar's G4 Storm waterblock with revised stepped base addition for LGA775 cpus
Swiftec MCP650 Pump
Thermochill 120.3 with 3x 120x38mm Panaflo 103cfm fans on 3.5" Rheobus 13W per channel
Dangerden fill port
Clearflex 1/2" ID tubing

Stock benchmarks:

Winbond Hardware doctor
http://www.fileshosts.com/DFI/875P_T...cpuz_idle1.jpg

Sandra 2005
http://www.fileshosts.com/DFI/875P_T....8-1.6/cpu.jpg

http://www.fileshosts.com/DFI/875P_T.../cpu-media.jpg

http://www.fileshosts.com/DFI/875P_T...6/mem-buff.jpg

http://www.fileshosts.com/DFI/875P_T...mem-unbuff.jpg

Everest Home
Read: 5,808 MB/s
Write: 2,056 MB/s
Latency: 80.3 ns

Super Pi unpatched
1M = 38s
2M = 1min 26s
8M = 6min 47s

Super Pi patched
1M = 34s
2M = 1min 19s
8M = 6min 18s

StressPrime 2004
16+ mins

Overclocking:

I still haven't found my way around the bios options and not sure how well this memory performs on the 875P-T yet but I was able to boot and take this screenie at stock vcore - not stable

http://www.fileshosts.com/DFI/875P_T...2-1.7/cpuz.jpg

Not bad temperature wise which was my main concern. I just have to test all my memory on this setup and see what works. I haven't installed my OCZ DDR Booster yet but I know it works with this motherboard so can pump more vdimm into my memory :)

So having a 875P chipset, does it also have the same affect in that 2x 512MB will result in more bandwidth than 2x 256MB memory modules ? I'll find out soon :D

Note: Seems the Winbond Hardware doctor's reported 12v rail is off off a bit, it shows 12.24-12.38v but from multimeter it says 11.98v idle and 11.94v load. So i'v raised it to 12.20v idle and see if that helps



.

Reserved for overclocked benchmarks

Full Pics here

http://www.fileshosts.com/DFI/875P_T...T_18_thumb.jpg http://www.fileshosts.com/DFI/875P_T...T_22_thumb.jpg http://www.fileshosts.com/DFI/875P_T...T_23_thumb.jpg http://www.fileshosts.com/DFI/875P_T...T_24_thumb.jpg http://www.fileshosts.com/DFI/875P_T...T_30_thumb.jpg

Nice. What r ur temp?
BTW I dont like screens @stock frequency :D

added some pics to 2nd post :)

This is where i am now priming

http://www.fileshosts.com/DFI/875P_T...essprime_2.jpg

i'm completing my full set of benchmarks @ 4097mhz and man is this 875P-T one screaming fast board.. think of Asus P4C800-E and Abit AI7 mixed into one, with fast stable cpu/system performance and the memory/GAT tweaks of AI7.

i.e. i'm able to pull off 3811MB/s memtest86+ v1.40 bandwidth @ 17x 241mhz 3-2-2-6 with 2x 512MB OCZ PC3700EB single sticks at 3.2v vdimm

Time for some overclocked benchmarks :) This 875P-T is very aggressive on memory and really puts out some nice memory bandwidth!

3.4E 550 SL7J8 @ 4097mhz at 1.5-1.53v vcore load
DFI Lanparty 875P-T 11/30 bios
2x 512MB OCZ PC3700EB Singles @ 241mhz 3-2-2-6 at 3.2v vdimm
80GB WD 800JB
580W Topower 6P686 Adjustable rail PSU

Sandra 2005
CPU = 11,997 / 4,909/8594
CPU Multimedia = 28,922 / 38,838
Buffered = 6,080 / 6,066
Unbuffered = 3,692 / 3,741

Everest Home
Read = 7,231 MB/s
Write = 2,632 MB/s
Latency = 66.4 ns

SuperPI
1M = 31s
2M = 1min 10s
4M = 2min 34s
8M = 5min 35s
16M = 12min 23s
32M = 27min 05s

SuperPI Patched
1M = 28s
2M = 1min 05s
4M = 2min 22s
8M = 5min 11s
16M = 11min 29s
32M = 25min 10s

Bloody fast for the low mem speed :D


.

That is bloody impressive. :D

thanks mate

Not stable but here's more benchies

3.4E 550 SL7J8 @ 4182mhz at 1.568-1.6v vcore load
DFI Lanparty 875P-T 11/30 bios
2x 512MB OCZ PC3700EB Singles @ 246mhz 3-2-2-6 at 3.2v vdimm
80GB WD 800JB
580W Topower 6P686 Adjustable rail PSU

Sandra 2005
CPU = 12,156 / 5,017/8,714
CPU Multimedia = 29,518 / 39,625
Buffered = 6,190 / 6,195
Unbuffered = 3,769 / 3,824

Everest Home
Read = 7,312
Write = 2,426
Latency = 65 ns

SuperPI
1M = 30s
2M = 1min 09s
4M = 2min 31s

SuperPI Patched
1M = 28s
2M = 1min 03s
4M = 2min 19s

dropping to 14x multiplier for some bandwidth benchies @ 251mhz 3-2-2-8 at 3.2v

Everest Home
Read = 7,343 MB/s
Write = 2,767 MB/s
Latency = 68.5 ns

Sandra 2005
Buffered = 6,239 / 6,239
Unbuffered = 3,686 / 3,747 (guess the lower 14x multipler reduced this considering i get higher @ 17x X 246mhz 3-2-2-6 here )

i fried two boosters on this board within seconds after first boot...

Quote:

---

Originally Posted by xsky

i fried two boosters on this board within seconds after first boot...

---

yeah i fried mine too at 3.2-3.3v too in same exact area that Fugger fried his :(

Quote:

---

Originally Posted by eva2000

yeah i fried mine too at 3.2-3.3v too in same exact area that Fugger fried his :(

http://www.fileshosts.com/DFI/875P_T...booster_01.JPG

---

welcome to the club :D

exactly that happened two times to me. my vdimm was set to 2.6V in bios and the booster died with 2.8V on display!!!!
it wasn't warm at all

was it fried in the same trace/area as above pic ?

yes. absolutely the same area as on the pic BOTH times

xsky, what psu and psu rail settings you had with the combo ?

first time: antec 550W True Power
second time: ocz powerstream 520W set to exactly 3.3/5/12V

well that seems to rule out the psu being the problem

just bridge the burn with a wire and the booster will come back to life...i did the same as you guys...LOL

I also wacked the 3.3V line up on the PSU and had 3.5Vdimm at the 3.2v setting which was nice...;)

oh you mean 3.3v rail upped without booster can give 3.5v ??? or only with booster in place ? got pics of the bridged trace ? i don't have the skill to do such a thing anyway... have to rma it once the retailer reopens heh

where do you measure vdimm without booster ? winbond hardware doctor on cd-rom displays vdimm but it beeps like crazy even at 2.8v vdimm, the newer winbond hardware doctor on web site at DFI removed vdimm display option

Tme you learn't how to solder there eva...i just bridged over the burn section to 2 solder points with a wire and the booster came back to life...real easy repair.Rear window defogger paint may do the same for you...apply it thick so it takes the current but the wire is a much better job.

No booster and 3.6v+ on 3.3v line upped the vdimm.

for those with burn't boosters from the 875P-T this should get the booster back up although it seems the booster is not 100% compatible with this board.

I would solder 3 wires to the 3 points from the 1 common point to spread the load. use wire that can carry 10A at 5V

http://www.xtremesystems.org/forums/...id=21550&stc=1

Quote:

---

Originally Posted by bigtoe

for those with burn't boosters from the 875P-T this should get the booster back up although it seems the booster is not 100% compatible with this board.

I would solder 3 wires to the 3 points from the 1 common point to spread the load. use wire that can carry 10A at 5V

http://www.xtremesystems.org/forums/...id=21550&stc=1

---

My booster didn't burn like the above pic but doesn't work anymore. Is there a work around to replace the adjustable pot or are the new units different. I'm not sure the pot is fried, but it worked for one day and just stopped being adjustable.

i can't even a tin a wire yet lol.. but i have 2 other boosters here so that's no prob for my other boards.. guess 875P-T doesn't like the booster... >3.3v psu rail for more vdimm sounds nice... all i need is to be able to see what vdimm i'm setting heh

Quote:

---

Originally Posted by mike.elmes

My booster didn't burn like the above pic but doesn't work anymore. Is there a work around to replace the adjustable pot or are the new units different. I'm not sure the pot is fried, but it worked for one day and just stopped being adjustable.

---

Just remove the pot and solder a 10k linear 20 turn or so cermet pot on...all 3 legs;)

dead easy and much more finely adjustable.

AndyT has a thread in the overclocking forum showing his modded unit.

hi!
can anyone show a pic of vdimm read point?
@bigtoe did you remember the "special" bios setting on the board already? ;)

eva - make sure your solder is rosen core, otherwise you will never get it to tin ;)

OCZ ought to beef up that area for future revisions.

idd or add it in the manual ;)

not sure what solder i was using i think it was 60/40 tin something heh

Okay swapped out my MX480EL card and put in my Powercolor X800XT PE for some fun at stock cpu speed :)

P4 3.4E 550 SL7J8 @ stock
Powercolor X800XT PE stock cooling
2x512MB OCZ PC3700EB Singles @200mhz 2.5-2-2-6

X800XTPE @520/560

3Dmark2001 = 19,855
3Dmark2003 = 13,298
3Dmark05 = 5,982
Aquamark3 = 70,543
PCMark04 = 5,510

X800XT PE @575/599

3Dmark2001 = 24,268
3Dmark2003 = 14,177
3Dmark05 = 6,435
Aquamark3 = 73,192
PCMark04 = 5,500

Hey Eva2000, do you know if it's possible to mount a vapo ls s775 kit on the DFI LP?

I think VapoLS fits all LGA775 boards.. but i only have a mach2 gathering dust and i know that doesn't fit heh

Well I truly went and broke my personal best score of 14,813 with my new 3.4E 550 SL7J8 and same Powercolor X800XT PE stock heatsink and with room to move too (no place to put the zalman fan bracket so no extra 190cfm delta air cooling over card yet).

But @ 575/602 still managed 3dmark2003 score of = 15,014

P4 3.4E 550 SL7J8 @ 17x247FSB = 4199mhz
DFI Lanparty 875P-T LGA775 board
Powercolor X800XT PE stock cooling @575/602
2x512MB OCZ PC3700EB Singles @247mhz 3-2-2-6

http://fileshosts.com/DFI/875P_T/res...2_15014_tn.jpg

Well last benchie for tonight

P4 3.4E 550 SL7J8 @ 17x248FSB = 4216mhz
Powercolor X800XT PE stock cooling
2x512MB OCZ PC3700EB Singles @248mhz 3-2-2-6

3Dmark2003 = 15,035
Aquamark3 @579/603 = 85,165

http://fileshosts.com/DFI/875P_T/res...2_15035_tn.jpg

http://fileshosts.com/DFI/875P_T/res...-603_85165.jpg

Broke my previous PCMark04 high score of 6,360 with my 3.2EE@245fsb = 3920mhz.

P4 3.4E 550 SL7J8 @ 17x245FSB = 4165mhz
Powercolor X800XT PE stock cooling @575/602
2x512MB OCZ PC3700EB Singles @245mhz 3-2-2-6 at 3.2v

PCMark04 = 6,526

http://fileshosts.com/DFI/875P_T/res...02_6526_tn.jpg

seems the clock generator on the board is a bit wacky

set FSB to
240 = 241 real
244 = 245 real
245 = 246 real
246 = 247 real
250 = 251 real

heh

woohoo Intel power baby!

Powercolor X800XT PE @604/604 = 15,343
http://service.futuremark.com/compare?2k3=3533043

http://fileshosts.com/DFI/875P_T/res...4_15343_tn.jpg

Powercolor X800XT PE @605/605 = 6,859 3dmark05
http://service.futuremark.com/compare?3dm05=434349

http://fileshosts.com/DFI/875P_T/res...05_6859_tn.jpg

Powercolor X800XT PE @605/605 = 85,426 Aquamark3
http://arc.aquamark3.com/arc/arc_vie...run=1023091253

http://fileshosts.com/DFI/875P_T/res...-604_85426.jpg

nice to see eva, any chance of Sandra buff+unbuff @ ~ 250FSB? just to compare to my 875P (i have FINALLY got full PAT working)

Quote:

---

Originally Posted by jem10

nice to see eva, any chance of Sandra buff+unbuff @ ~ 250FSB? just to compare to my 875P (i have FINALLY got full PAT working)

---

post #8 ;)

Quote:

---

Originally Posted by bigtoe

for those with burn't boosters from the 875P-T this should get the booster back up although it seems the booster is not 100% compatible with this board.

I would solder 3 wires to the 3 points from the 1 common point to spread the load. use wire that can carry 10A at 5V

http://www.xtremesystems.org/forums/...id=21550&stc=1

---

Exactly the same place my friends booster blew.

Soldering the 3 wires on when its complety working (no burnt marks) would it make it short? Would this be best to do this little mod before the booster touches the vdimm slots on the DFI?

Thanks...

should be fine to do that to help the traces handle more current.

You can run a connectivity test with a DMM to be sure but its pretty easy to see the layout of the traces ;)

do u have fluctuations of VCORE during the clock ..because my DFI 875P_T fluctuate a lot and I do not know reacon ..and did u do DROOP mod

No droop mod

yeah droops about 0.03 - 0.06v under load on vcore - apparently it will droop more if you're not using 24pin PSU

as to fluctuations i think it's just the buggy sensor :(

Hello Guys,
Sorry for the little off-topic reply- Eva2000!

Quote:

---

I think VapoLS fits all LGA775 boards.. but i only have a mach2 gathering dust and i know that doesn't fit heh

---

In my opinion, the LGA adapter by Asetek is not as good as the socket478/754 unit, especially on the DFI 875P-T mobo. 1st off, the capacitors on the top-side of the cpu socket sits way to close to the socket, there is a lot of condensation happening here! Enough to fry your board.

It is not that I am a noob with Vapochills, I have been using them for the past 2.5 years, but this current socket combined with the bad capacitor layout on this specific board is looking for trouble. Rather get the 865 chipset LGA version of this board, I see the have moved the caps further away from the socket ;)

If you need links, I can post them.

Br,
MrBean.

Thanks for the info MrBean, it's actually on topic since i've been tempted by the VapoLS but with a mach2 already was deciding... guess LGA775 board layouts are really messing with current cooling products compatibility!

I had trouble installing the vapo on 875p-t, too.
Used some additional armaflex...the coldest temperature I could messure around the kit was about 17°.

Quote:

---

Originally Posted by MrBean

In my opinion, the LGA adapter by Asetek is not as good as the socket478/754 unit, especially on the DFI 875P-T mobo. 1st off, the capacitors on the top-side of the cpu socket sits way to close to the socket, there is a lot of condensation happening here! Enough to fry your board.

It is not that I am a noob with Vapochills, I have been using them for the past 2.5 years, but this current socket combined with the bad capacitor layout on this specific board is looking for trouble. Rather get the 865 chipset LGA version of this board, I see the have moved the caps further away from the socket ;)

If you need links, I can post them.

---

Wow, very informative post Mr. Bean. I was just about to pull the trigger on this combo and the 775 kit. But what about the evap heating element of the Vapo? What do you set it at?

I thought my P4P800 had caps close to the socket, and they dont get wet or cold. But looks like the DFI board has them actually inline with the screws/mounting bracket.:eek:

http://209.35.180.29/emotes/775.jpg

Vapo/875T pics would be awesome, please do! :up:

Quote:

---

Originally Posted by xsky

I had trouble installing the vapo on 875p-t, too.
Used some additional armaflex...the coldest temperature I could messure around the kit was about 17°.

---

Do you get condensation? Do you use the Vapo's heating element?

not sure about Vapo but here's what one guy did to mod Mach2 GT P4 kit for LGA775 http://www.extremeprometeia.com/foru...0230#post10230

I didn't notice any condensation until now and I checked carefully.
Heater in Vapo Head @ 40% and 30% on the MB back

Another option is to just buy a 775 kit for the Prommy. I believe they are available now.

Bump.

I went down and bought me a 530E ~ 3.0 proc. I was really curious about if the AGP/PCI lock was just a 14 multi issue or if it effected the default multi setting. I also was wondering of the 5:4 divider issue was at just the 14 multi setting or at the default multi also. And I was also concerned about comments made that memory might not be able to OC as high by would still throw high bandwidth numbers.

The proc I got is a SL7PU ~ 3447A608 ~ Costa Rica

You can see from the numbers I did not have a PCI Lock failure at least up to 277fsb.

I could not go any higher because 277fsb is as high as my 3200 Plat Rev2 will go. That is the same max I got out of them when I had them in my IC7 Max3. The following benchmarks are pretty much the same as I got when they were in my Max3.

The 5:4 divider still does not work. But these OCZ3200 never have been good with the 5:4 divider. I'm going to try my Adata in a couple of days.

Cpuz
http://home.comcast.net/~stanspam/Public/CPUz.JPG

Sanda Cpu
http://home.comcast.net/~stanspam/Public/SanCPU.JPG

Sandra UnBuffered
http://home.comcast.net/~stanspam/Public/SanUB.JPG

Sandra Buffered
http://home.comcast.net/~stanspam/Public/SanBuff.JPG ~ Sandra Mem Buff

Super Pi (hacked)
1 MB
2 MB
4 MB

Super Pi Regular
1 MB
2 MB
4 MB

interesting investigation there Bigstan :)

did you try that thing with the vdimm....upping the 3.3V rail.

I still don't know where to read vdimm

tried upping 3.3v rail but didn't help with my mem past 3.2v vdimm :(

Just playing with some raid0 tests.

3.4E @ 3415mhz stock 1.3875v
DFI 875P-T 12/30 bios
64MB Albatron MX480EL
2x 512MB OCZ PC3700EB singles @200mhz 3-2-2-6

2x 250GB Maxtor Maxline II Plus 8MB Cache PATA drives using Abit Serillel2 adaptors to connect to board's SATA 1 + 2 ports.

Raid 0 using 128k stripe and 16k clusters

ATTO

http://www.fileshosts.com/DFI/875P_T...k/ATTO_4mb.jpg http://www.fileshosts.com/DFI/875P_T.../ATTO_32mb.jpg

Drivebench

http://www.fileshosts.com/DFI/875P_T...ebench_4MB.jpg

http://www.fileshosts.com/DFI/875P_T...bench_32MB.jpg

http://www.fileshosts.com/DFI/875P_T...ench_128MB.jpg

Sandra 2005 Lite

http://www.fileshosts.com/DFI/875P_T...ilesystems.jpg

Now interesting is when i ghost imaged my existing OS drive to a empty raid0 array - where existing partitions such as C (sized to 81GB), uses default 4k cluster size, this is what i get

http://www.fileshosts.com/DFI/875P_T...OSimaged_C.jpg http://www.fileshosts.com/DFI/875P_T...OSimaged_C.jpg

http://www.fileshosts.com/DFI/875P_T...OSimaged_C.jpg

http://www.fileshosts.com/DFI/875P_T...OSimaged_C.jpg

http://www.fileshosts.com/DFI/875P_T...OSimaged_C.jpg

downloaded hdtach3 free edition and did

short test 8MB
http://www.fileshosts.com/DFI/875P_T...OSimaged_1.jpg

long test 32MB
http://www.fileshosts.com/DFI/875P_T...ged_32mb_0.jpg

compared to single 250GB Maxtor Maxline II Plus PATA drive
http://www.fileshosts.com/DFI/875P_T...ine_single.jpg

I doubt upping the 3.3V rail gives me more vdimm. There was not 1MHz difference between 3.3V and 3.7V with BH6!

Decided to have some extra fast fun and I loaded up RamDiskXP Pro for a few runs

http://www.fileshosts.com/DFI/875P_T...amdiskXP_1.jpg

http://www.fileshosts.com/DFI/875P_T.../ATTO_32MB.jpg http://www.fileshosts.com/DFI/875P_T...B_241_3226.jpg

http://www.fileshosts.com/DFI/875P_T...filesystem.jpg

http://www.fileshosts.com/DFI/875P_T...m_241_3226.jpg

http://www.fileshosts.com/DFI/875P_T...2005_flash.jpg

http://www.fileshosts.com/DFI/875P_T...h_241_3226.jpg

Too bad real HDs can't run that fast!