1366 x58 Xeon Enthusiast overclocks club

Fuck it, I'm going to delid my CPU and run it without a spreader at all, it will probably take me the same amount of time to lap the sucker, but the delid will be less physical labour and will result in lower temps.

I've got 2 more CPUs just lying around so I'm going to use one as a trial run and the other can be a spare in case I completely fuck it up.

Help me pick the guinea pig:
i7 930 (45nm 130W 8MB Cache) or
X5677 ES (32nm 130W 12MB Cache)

Both will do 4.2Ghz
 
Deimos said:
Fuck it, I'm going to delid my CPU and run it without a spreader at all, it will probably take me the same amount of time to lap the sucker, but the delid will be less physical labour and will result in lower temps.

I've got 2 more CPUs just lying around so I'm going to use one as a trial run and the other can be a spare in case I completely fuck it up.

Help me pick the guinea pig:
i7 930 (45nm 130W 8MB Cache) or
X5677 ES (32nm 130W 12MB Cache)

Both will do 4.2Ghz
Click to expand...


You can not delid a socket 1366 CPU, the heat sperader is soldered to the core...

With the same settings the i7 930 should be a little faster but not more than ~0.5%.
Strange that the X5677 ES do not clock higher.

I would use the X5677 as long as it runs cooler and can run @4.2 with lower voltages.
 
Last edited:
Deimos said:
Fuck it, I'm going to delid my CPU and run it without a spreader at all, it will probably take me the same amount of time to lap the sucker, but the delid will be less physical labour and will result in lower temps.

I've got 2 more CPUs just lying around so I'm going to use one as a trial run and the other can be a spare in case I completely fuck it up.

Help me pick the guinea pig:
i7 930 (45nm 130W 8MB Cache) or
X5677 ES (32nm 130W 12MB Cache)

Both will do 4.2Ghz
Click to expand...

You can't delid these, but you can buy ones without the IHS from a mac pro.
 
Deimos said:
Fuck it, I'm going to delid my CPU and run it without a spreader at all, it will probably take me the same amount of time to lap the sucker, but the delid will be less physical labour and will result in lower temps.

I've got 2 more CPUs just lying around so I'm going to use one as a trial run and the other can be a spare in case I completely fuck it up.

Help me pick the guinea pig:
i7 930 (45nm 130W 8MB Cache) or
X5677 ES (32nm 130W 12MB Cache)

Both will do 4.2Ghz
Click to expand...

LOL. How did that de-lid go?

(please tell me you read the last two comments before attempting this operation)
 
I'm well aware that the heat spreader is soldiered on, it doesn't mean you can't delid them, there are plenty of guides around.

So which one is my guinea pig?
 
Went with the X5677.

Nerve racking :eek:

Took me about 10 minutes, I cut the silicone around the edge, then I used the stove top method, dropped it on the element and after just a few seconds heard a "pop" and it was off, just have to clean the solder off the core now and it's good to go, I might pull out my soldering iron or heat gun and suck the solder off, I have temp regulated equipment so should be able to do it without causing any damage.

If anyone wants to attempt this, a quick tip: use two blades on both sides of the heat spreader when you heat it up, that way the CPU will come off well clear and the solder wont set and stick back to the heat spreader, and there is a lot of it...
 
Last edited:
Deimos said:
Went with the X5677.

Nerve racking :eek:

Took me about 10 minutes, I cut the silicone around the edge but I'm wondering just how necessary it is to go all the way around, I'm thinking that just one corner may be enough just to get some leverage for when you heat it up, the silicone seems to let go from the heat as well.

I used the stove top method, dropped it on the element and after just a few seconds heard a "pop" and it was off, just have to clean the soldier off the core now and it's good to go, I might pull out my soldiering iron or heat gun and suck the soldier off, I have temp regulated equipment so should be able to do it without causing any damage.
Click to expand...

impressive.....well done....i would not have the balls to try it...but hell lets us know if it really helps in the temps...im very curious:eek: oh and throw up some pics.....really
 
This is just a trial run before I delid my X5670 so the most important thing right now is just to see if the CPU still works.

I'm going to delid the 930 as well because I'm not 100% confident in the method I used just yet and I don't want to jump in and do the X5670 without 100% confidence.
 
Just for you guys I'm going to post detailed instructions with exact measurements of the silicone bead. I can already see that I may have cut in to a couple of the surface mount caps on the cpu (but only just so it may still be OK) but I would suggest measuring and drawing a line on the blade before beginning so you know exactly how far you need to cut.

I'm going to test the silicone to see if it also melts at high temp as well just to figure out if it is necessary to cut it all or just a corner.
 
First attempt has failed unfortunately.

Result:

Boots in to windows at stock clock but 8GB of ram is missing sadly. It might be possible to repair the CPU but I don't have the skill to pull it off.

Temps are quite remarkable, 45C right now while running prime with a 6C spead across all cores, keep in mind at stock clocks this CPU is 130W.

I'm still not keen to do the X5670 until I have had at least one successful delid so I'm going to attempt the i7 930 right now, I'm pretty sure I can do the next one as I know where I went wrong the first time round.

Will post the results when I'm done.

I'll do a video for this one.
 
Well fuck, the second attempt was a miserable complete failure.

The worst part is I'm not quite sure what I did that killed the 930 :( it came off really cleanly, I didn't scratch the substrate or get any where near the surface mount caps, the CPU looks pristine.

I went back to the X5677 and attempted to resolder the damaged caps and it is now detecting 8GB of ram so yay, there is hope yet that I can get this thing fully operational.

Unfortunately overclocking is off limits when the ram is not fully working because the motherboard assumes its a bad OC and reverts back to stock clocks.

Temps are even lower now as I have switched to NT-H1, the temps seem impossible, one of the cores is reading below ambient now, WTF!?
 
Last edited:
HOLY SHIT! :eek:

I bit the bullet and did the X5670.

Turns out my first two attempts were successful :eek:

The problem turned out to be the pressure from the waterblock wasn't enough, I had to remove the plastic standoffs (shorter ones I found just for this occasion) and tighted the screws quite a lot more to push the cpu in to the pins more, once I had done that the PC booted with all ram showing on the X5677!!

I'm running the X5670 right now with prime and the temps are at least 10C lower.
 
Deimos said:
HOLY SHIT! :eek:

I bit the bullet and did the X5670.

Turns out my first two attempts were successful :eek:

The problem turned out to be the pressure from the waterblock wasn't enough, I had to remove the plastic standoffs (shorter ones I found just for this occasion) and tighted the screws quite a lot more to push the cpu in to the pins more, once I had done that the PC booted with all ram showing on the X5677!!

I'm running the X5670 right now with prime and the temps are at least 10C lower.
Click to expand...

Dude, nice!

I for one will never join you in this endeavor. Bare die may be more efficient, but I remember the days of strapping down HSF's to bare cores on Tbirds and the sound of a die crunching under the block once was enough for me. Only happened once, and I was putting that thing on as properly as could be done, still, *Crush*.
 
Im very impressed with your results as well....but i never could get heat sinks to mount correctly back in the old days...i guess i could but results were all over the place...since i been using cpus with heat spreaders i have never had a problem once....its so fucking easy to install heat sinks now...with that being said i don't think i would risk it either....but keep us informed how its working out...lower temps has to be nice;)

are you using core temp to monitor the temps? post some screenshots after Intel burn test running max at 5 tests or something......im just curious were we stand one against the other..also don't forget to post pics of your cpu delided ...never saw one before:)
 
Last edited:
After testing for a little while I have noticed that one of my monitoring programs has dropped 10C but IXTU and HWMonitor are 9-10C higher.

Current settings results in a max temp of 70C after an hour of prime:
1.36V Core
1.88V PLL
1.25 IOH
1.64 Mem
200 BCLK
22 Multi (4.4Ghz)

with fans and pump at 50% (almost silent)

I made a change to my loop recently so those temps may improve once the air is out of the loop but I'm having some difficulty on that front, if I run the pump at max it seems to churn out a lot more bubbles and I don't know if it is caused by the new EKWB-Bay RES D5 or if it is air still in the loop.

I'll post pics soon, right now I'm keen to get some time in with Watch_dogs.
 
I just got my Xeon x5670 running at 4.4GHz without too much tweaking.

1.28V Core
200 BCLK
22 x Multiplier
Air cooled with Thermal Right IFX14 heatsink.
30 GB of DDR 1600 Memory ( 3 x 8GB, 3 x 2 GB)
Asus P6T v2 Motherboard
 
evilmedic said:
I just got my Xeon x5670 running at 4.4GHz without too much tweaking.

1.28V Core
200 BCLK
22 x Multiplier
Air cooled with Thermal Right IFX14 heatsink.
30 GB of DDR 1600 Memory ( 3 x 8GB, 3 x 2 GB)
Asus P6T v2 Motherboard
Click to expand...


Your CPU is even better than mine!
My X5670 needs 1.30v Core to be long term Prime stable at 4.4 with 6 sticks of 1600 MHz memory. 21x210 or 22x200 makes no difference the CPU needs the same 1.30v.

Here is a screenshot of my X5670 priming at 4.4. The temperature sensors are a bit dodgy but the one that shows the highest value is more or less accurate.

http://u.cubeupload.com/jonaz81a/Prime2474400.jpg
 
Skyscraper said:
Your CPU is even better than mine!
My X5670 needs 1.30v Core to be long term Prime stable at 4.4 with 6 sticks of 1600 MHz memory. 21x210 or 22x200 makes no difference the CPU needs the same 1.30v.

Here is a screenshot of my X5670 priming at 4.4. The temperature sensors are a bit dodgy but the one that shows the highest value is more or less accurate.

http://u.cubeupload.com/jonaz81a/Prime2474400.jpg
Click to expand...

had to back mine down to 4200mhz...just couldn't find the magic settings to get mine 100% could be any number of factor's like running certain voltages to high.....fuck it...4200hz will have to do
 
I have been lazy lately and the rebuild of the dual X5650 on my SR-2 has moved forward slowly (aka not at all). Today I decided not water cool that build at this moment although I have all the parts. I will throw the build together with the Noctua D14s for some fun and play.



First an image of the build as it were. Dust, more dust and even more dust.
Image00002.jpg




The CPUs! X5650 + X5650. 22x multiplier here I come! (Before the rebuild the CPUs were X5670 + X5650 so I was limited to the 20x multiplier)
Image00003.jpg




The board and some memory and some extra CPUs and some extra SR-2 and... dont hate me.
Image00001.jpg
 
omg....thats a crime...clean them things off lol;)....if you never tried it a shopvac on blow it works great for monthly blow outs on mine....keeps its almost brand new clean....and all ya have to do is pull the side off...im done cleaning in 5 min tops...takes more time to plug all the cables back in....you don't want to do it inside your house, unless you just love dust....you be amazed the dust cloud a shop vac creates during the monthly blow out
 
primetime said:
omg....thats a crime...clean them things off lol;)....if you never tried it a shopvac on blow it works great for monthly blow outs on mine....keeps its almost brand new clean....and all ya have to do is pull the side off...im done cleaning in 5 min tops...takes more time to plug all the cables back in....you don't want to do it inside your house, unless you just love dust....you be amazed the dust cloud a shop vac creates during the monthly blow out
Click to expand...

During the almost 4 years I have had the SR-2 that build has been cleened only once. That was when the CPUs got changed from 4 core CPUs to 6 core CPUs half a year after I first put the build together.
The spare SR-2 is new, I did only test it once 3 years ago. The E5620 CPU, the two sticks of memory and even the cooler I used for the test is still mounted to the board... lazy...

Dirty heatsinks are not good for overclocking though so the Noctua D14s got a really good cleaning.

The machine is running nicly now. I posted a score in the Cinebench thread.
22*200 = 4400 MHz with 1.4V Core voltage and 1.3V VTT for 3600MHz Uncore (1.4V VTT for 4000 MHz Uncore). The memory run nicly at 2000 MHz.
 
I've backed mine off to 4.2 for the moment, even after delidding the CPU was hitting 70C under prime and I want my PC to be virtually silent.

I backed my water pump way off, its an MCP655, I've set it to 2 (out of 5) and used the CPU PWM to control almost all the fans, it's currently quieter than my HTPC with idle temps below 30C

I made a cable mod so that I can run all the fans off the PSU while maintaining PWM control off the motherboard, I managed to pull it off without permanently modding any cables while using just one tool - a small screw driver.

I'll post pics later today if I get a chance.
 
looks great...post all the pics you case,,,,,,,,,possibly one on 2 of night join the process
make a sweet guide and others will follow;)
 
Here it is, full specs.
Case and water cooling:
Silverstone TJ07
Swiftech Apogee waterblock
Swiftech MCR655
EKWB Bay Res D5 pump housing/reservoir
Feser X-Changer 360 radiator

Hardware:
SeaSonic Platinum-1000 PSU
Xeon X5670 @ 4.3Ghz (21x205 1.28V)
Asus P6X58D Premium
12GB G.Skill RipJaws
Sapphire Tri-X R9 290X 4GB GDDR5 OC
Soundblaster Zx
Samsung 840 Pro 512GB (OS)
Samsing 840 256GB x 2 (other shit)

3 x Dell U2412M
Logitech z5500 speakers

Original configuration had 2 x 500GB Western Digital RE, an i7 930 at 4.2Ghz, an X-Fi Titanium and a Radeon 5870

IMG_1528.jpg


IMG_1516.jpg


IMG_1527.jpg


IMG_1524.jpg


IMG_1519.jpg
 
I cant really be bothered to put the SR-2 system back in its case, I have to tear it apart when switching to water so there is no point.



Only the board with the two Noctua D14s.
ddImage00001.jpg



The whole system. The GTX 460 is a place holder but its enough for Blizzard games.
163Image00002.jpg
 
I thought about getting an SR-2 and another 5650, but I didn't want to have to mod my case to fit the board.
 
I think I've found the sweet spot for my CPU, 4.3Ghz (205BCLK x 21) with a tiny bump in volts from 4.2 (+.03v).

I'm at 10 hours of prime with just 1.288v on the core. PLL is at auto and IOH is at 1.26v

I'm still tweaking it though, I have a feeling that I may be able to go higher still but instead of increasing core voltage I need to start increasing the IOH and PLL volts, even after giving it 1.36v at 4.4Ghz it wasn't stable until I increased PLL and IOH so it may still run at 4.4Ghz with lower core volts.
 
Last edited:
Testing new settings right now with IBT:
21x209BCLK (4384 Mhz)
1.296v
1.8v CPU PLL
1.3v IOH
1.7v Ram

Primetime maybe you should try increasing IOH instead of CPU volts to get to 4.4Ghz, I'll keep you updated if mine works out.
 
Deimos said:
Testing new settings right now with IBT:
21x209BCLK (4384 Mhz)
1.296v
1.8v CPU PLL
1.3v IOH
1.7v Ram

Primetime maybe you should try increasing IOH instead of CPU volts to get to 4.4Ghz, I'll keep you updated if mine works out.
Click to expand...

IOH huh......Don't think i ever thought of boosting that.....let me know how it works out...things is..... its bullet proof right now...no failed test, no overheating, no crashes, nothing odd at all
And the benchmarks haven't shown a huge gain form 4200-4400 mhz...almost like it didn't matter.....1.7v on the ram huh? And I thought i was pushing it at 1.66 lol

im running 1.35 on the v core and 2 volts on pll...i been told its dangerous as fuck but i think if you look for the tell tell signs.....crashes, instability, overheating and you experience none of these ever...then its ok.....i will not be upping those no matter what since there already pushed pretty hard...but yea keep me posted how the ioh voltage works....never thought of that
 
primetime said:
IOH huh......Don't think i ever thought of boosting that.....let me know how it works out...things is..... its bullet proof right now...no failed test, no overheating, no crashes, nothing odd at all
And the benchmarks haven't shown a huge gain form 4200-4400 mhz...almost like it didn't matter.....1.7v on the ram huh? And I thought i was pushing it at 1.66 lol

im running 1.35 on the v core and 2 volts on pll...i been told its dangerous as fuck but i think if you look for the tell tell signs.....crashes, instability, overheating and you experience none of these ever...then its ok.....i will not be upping those no matter what since there already pushed pretty hard...but yea keep me posted how the ioh voltage works....never thought of that
Click to expand...

Have you tried default PLL? (1.8v) I haven't increased mine at all, just did 10 passes of IBT with those settings, gonna try prime now but so far so good.

Try drop your core volts a bit and compensate with a bit more IOH, see how you go, I was able to achieve 4.2Ghz with 1.25v without touching any other voltages (before delidding).

I have been told that as long as your ram volts are within .5v of your IOH that you are safe so if you want more ram volts, increase the IOH
 
Deimos said:
Have you tried default PLL? (1.8v) I haven't increased mine at all, just did 10 passes of IBT with those settings, gonna try prime now but so far so good.

Try drop your core volts a bit and compensate with a bit more IOH, see how you go, I was able to achieve 4.2Ghz with 1.25v without touching any other voltages (before delidding).

I have been told that as long as your ram volts are within .5v of your IOH that you are safe so if you want more ram volts, increase the IOH
Click to expand...


On the EVGA SR-2 I use 1.5v PLL (1.8v) for both CPUs.
When running 6*2 GB old Corsair memory I could run all voltages (except Vcore, VTT and memory) at the lowest setting even at 200 BCLK but now with 6*4 GB Samsung registered memory I had to up the IOH from 1.1v to 1.2v, otherwise sound would be distorted over 196 BCLK.

Its a shame that the Asus boards dosnt support lower PLL (1.8v) than default.
 
Is there a list somewhere of motherboards that are known to support the X56** Xeon's?

I have access to an DFI brand X58 mobo and am unsure about its compatibility. The bios notes on the website says it supports "Gulftown", but the X56** Xeons are "Westermere", yeah?
 
chameleoneel said:
Is there a list somewhere of motherboards that are known to support the X56** Xeon's?

I have access to an DFI brand X58 mobo and am unsure about its compatibility. The bios notes on the website says it supports "Gulftown", but the X56** Xeons are "Westermere", yeah?
Click to expand...

Yes they are.
 
chameleoneel said:
right, ok. But I'm still needing a list of known compatible boards. Because it seems most Asus boards will work----yet their bios notes don't mention Westmere at all.
Click to expand...

The BIOS notes may not but the product page certainly does.
 
You must log in or register to reply here.
Back
Top