1366 x58 Xeon Enthusiast overclocks club

[primetime]

Edit:
I wanted to include this in the first post since the information is super important as so people when trying to find there max overclocks they don't damage there brand new cpus. Please do not exceed these values as a safety precaution if you don't wan't to fry your brand new 32nm cpu

Zoson [H]ardness Supreme, 13.1 Years

vCore: 1.45v
CPU PLL: 1.9v
QPI/UCLK/Vtt: 1.35v
vDDR: 1.65v (Possibly QPI+0.5v)
IOH/ICH: 1.25v
PCI-E: 1.51v

Here's the datasheet for reference
http://www.intel.com/content/dam/www...heet-vol-1.pdf

Edit#2 Everyone be aware of how easy it is to bend pins on the motherboard if its done in a rush or not placed down on the side when doing the swap. Bent pins will rune your board and cause you to have a really bad day. Take your time and be careful and or you could be shopping for a new board before you even get started.

Wanted to start a new thread dedicated to using x58 1366 platform and using the popular used xeon 6 core cpu's going around everywhere ...back in the day we paid over 300 dollars for 920 cpus but now a days were flying with 44000mhz 6 core cpus sometimes at prices under 100 bucks a pop. Thats an unbelievable price per performance in my opinion

The last thread i started has over 65000 views and takes quite a wile to read threw. Anybody still running a socket 1366 motherboard might be better off just throwing in a hexa core cpu and overclocking it versus spending several hundred dollars upgrading the hole thing system....some of us love making the most of what we have...cause thats all we have.
This thread is not to brag about how bad ass our system is, but rather share info helping each other getting the most for there money...and get there systems running best as possible with least amount of money spent.

Thees still lots of people unaware there x58 system STILL has plenty of performance and yeas ahead of it....I be willing to bet our systems still overpower amd and or intel setups 5-6 years newer...i could be wrong but i think the benchmarks could prove it....
I'm hoping to get lots of people posting there success stories or problems with as many screen shots as possible proving stability/ performance for our sometimes 6 year old systems .....This is my day day out clocks....it just works

[/URL][/IMG]

[/URL][/IMG]


Edited to include first of many Deimos contributions

Some nice modded Bioses
http://www.bios-mods.com/

P6X58D Premium
P6X58D-E
P6T Deluxe (perfect for me)
speed test after using the newest (Updated Modded) for my board
Ill post more links in time as they come it

[/URL][/IMG]

[/URL][/IMG]


Edit for new content:


For eVGA X58 Classified (3way) rev 1.0 boards that don't support Westmere-EP Hexa Cores
ChineseStunna posted some info on modding it so it can work with these

http://forums.evga.com/x58-classifie...-m2211553.aspx
http://forums.evga.com/132BLE758A1-R...-m2153248.aspx

or this from tbob22
http://www.overclock.net/t/1461359/...eon-l5639-benchmarks-inside/290#post_21897471

 

[primetime]

hyper_pi_0.99 results

[/URL][/IMG]

 

[W.Feather]

Moved at request of OP to Intel Processor Sub Section

 

[Deimos]

Delid experiment on an X5670

Read fully before attempting this, my first recommendation is don't try it unless you like throwing money away.

This is the naked X5670.

Steps and tips:

If anyone wants to give this a go it's actually pretty straight forward, here are my recommendations:

• Use a short box cutter blade (the type that is shaped like a triangle with a corner cut off) and use a pair of locking pliers to hold it, this gives you way more control over the blade and how much force to use when cutting the silicone, it will also save your fingers from pain.
• When cutting the silicone, press and wiggle the blade back and forth its much cleaner and easier this way, don't try to cut it by dragging the blade around the edge.
• Use only one blade
• Put your blade in to one corner, then use a pre-heated stove element to heat the spreader to melt the solder, the CPU will "pop" off (you will actually hear a "pop") leave the spreader on the element and lift off the CPU, slide the spreader off the element only after the CPU is removed otherwise it will just set again and you will have to heat it once more to remove it.
• Heat the remaining solder and wipe it off with something rather than try to cut it off with a blade, finish off with a polishing cloth to give it a good flat finish.

Further notes:

You must remove the retention clip otherwise your heatsink or waterblock will fail to make contact with the CPU core. There are two possible ways to do this:
The easiest way is to remove the entire bracket by unscrewing it.
Another, possibly better way is to remove the top part of the mechanism and lever, leaving the back plate and surround still on the board, this has the benefit of reducing the amount the motherboard flexes when pressure is applied to the socket. If you decide to go this route you need to make sure the HSF/WB sits completely inside the retention bracket (mine does with room to spare).

When screwing on the HSF/WB it becomes very difficult to screw it on flat, don't be too concerned, just don't screw it on too tight or you will kill the CPU like I did
As you screw the HSF/WB on get a torch and check the clearance of the HSF/WB, it should be around half a mm clearance between each edge of the socket and your HSF/WB, not enough pressure and the pins won't make adequate contact with the CPU, you may not get post or have only some of your RAM show up if you don't have enough pressure on the CPU.

Final note: delid isn't worth the risk IMO, I killed a Xeon and an i7 930 from suspected over tightening of my WB, the CPU worked fine for quite a while but when I swapped out the motherboard I was careless and killed a couple of CPUs all too easily. The temp difference is marginal, my first impression was that there was a significant drop, and have seen others report that they can run lower volts on delided CPUs, however in my own testing on this platform where the lid is soldered on, I can't say for sure that there was any significant drop in temps and it made no difference to my overclocking.

Asus P6X58D Premium + LSI 9211 RAID card

The LSI 9211 can give a nice improvement for SSDs, even without raid the card will outperform the Marvel controller by 20% in maximum transfer speeds and 30% better access letency (this isn't really noticeable though as the access latency is in nanoseconds)

The only caveats with this config is the boot times are glacial and the card will only work in one of the top two PCIe x16 slots (the bottom slot caused issues for me).

The LSI also has lower access latency and CPU usage.

Single disk performance

Update 05/08/2014 - I can only seem to get my LSI 9211 to work in the top slot on both the Asus P6X58D Premium and the Asus Rampage III Extreme

RAID 0 performance

Some interesting results running RAID 0 on the built in controllers.

I ran the same benchmark on the LSI with RAID 0 but for some reason I lost the screen shot. For some reason the LSI controller suffers in small file transfers when set up with a RAID 0 array, The ICH10R and even the Marvell bests the LSI controller in small file transfers. once file sizes go over 32MB though the LSI controller takes the lead by a wide margin, achieving a max transfer speed of around 1GB/s

And this is the original configuration that I abandoned a while ago, its a 6 disk RAID 0 array with OCZ Vertex 3 120GB drives. As you can see, small file transfer performance is awful. The array is also pretty slow, the hardware is capable of around 3GB/s but going from 2 disks to 6 only increases max speed by 300MB/s

X58 24GB RAM limit

As far as I can tell this limit is only for the consumer CPUs (i7s).
Intel ARK lists the i7s as having a 24GB limit but the Xeons are listed as having a 288GB limit.

I can confirm that 48GB (6 x 8GB G.Skill RipjawsX) works on a Rampage III Extreme and the Asus P6X58D Premium.
Total ram is useable during my testing, I used IBT on Windows 8.1 x64 and told it to use 40GB of ram and it happily filled it. Sisoft Sandra was also quite capable of filling it up.
I have not personally noticed any difference with stability with the larger amount of ram with the same overclock.

Memory overclocking

In my own testing, the Xeons cannot overclock the ram without increasing BCLK, this is different from the i7 which allow ram overclocking without bumping BCLK. You must run with your ram at 1333 at stock clocks otherwise no POST.

Also, overclocking memory results in higher temps on the CPU which may limit your CPU overclock. I wouldn't recommend going far beyond 1600mhz, even just a small increase of 150mhz over this bumped my temps by up to 9C. YMMV.

There is another thread on this topic for the i7 4770k, same thing applies here.

X5660 vs newer X79 based hardware

There is a nice post over at overclock.net with in-depth comparisons between the X5660 vs newer systems, including a few gaming benchmarks with surprising results.

[Official] - Xeon X5660-X58 Full Review, Discussion & Comparison to X79 High-End CPUs [and Xeon L5639 benchmarks inside] - Longest Post Ever!!!

 

[primetime]

I've just been posting about some issues I'm having with my setup, I'm taking a break from tweaking for a little while, give this new CPU some time to settle I guess.

Currently running the same (X5670) @ 4.03 Ghz (24x168 BCLK) with Ram at 1:1 and 1.256V
Temps are amazing.

I've had this system pass 24h Prime at 4.4Ghz (22x200 1.35V) but for some reason my X-Fi Titanium doesn't like it. Haven't had time to continue trying to resolve it, at this point I'm considering upgrading to a newer sound card.

I haven't actually tested the stability of my current overclock (4.03Ghz) because I'm desperate to just play some games (with sound LOL) especially Watch_dogs, I have very limited alone time at the moment and frankly I would rather play games than spend time tweaking.

So is your sound working fine at the lower clocks? Yea that could make me crazy as well...luckily the sound blaster Z has has shown no problems here.
It seemed like i was passing 9 out of 10 tests at 4400mhz but maybe it just wasn't meant to be idk.......im going to focus on getting it stable a 4200mhz in the mean time. 21x instead of 22x.........as much as i hate to do so

 

[Deimos]

I thought my sound was working at the lower clock but I loaded watch dogs last night and the sound cut out after a couple of minutes.

It turns out it was the PSU, 3.3V was only at 3V, I swapped out the PSU and all is fine now at 4.4Ghz...

 

[primetime]

I thought my sound was working at the lower clock but I loaded watch dogs last night and the sound cut out after a couple of minutes.

It turns out it was the PSU, 3.3V was only at 3V, I swapped out the PSU and all is fine now at 4.4Ghz...

awesome news...great troubleshooting there!

 

[primetime]

back to 4400mhz...gonna keep at it as long as it takes

[/URL][/IMG]

 

[Deimos]

I heard that IBT is really no good for stability testing, even prime95 apparently. there is some other testing suite that will test every part of the CPU instead of just the narrow band of functions that IBT and P95 test, I forget what software it is though.

Someone recommended the stress test built in to SiSoft Sandra...

 

[primetime]

I heard that IBT is really no good for stability testing, even prime95 apparently. there is some other testing suite that will test every part of the CPU instead of just the narrow band of functions that IBT and P95 test, I forget what software it is though.

Someone recommended the stress test built in to SiSoft Sandra...

going to give occt an hour to see if it finds any errors..and update results

 

[ccityinstaller]

So is your sound working fine at the lower clocks? Yea that could make me crazy as well...luckily the sound blaster Z has has shown no problems here.
It seemed like i was passing 9 out of 10 tests at 4400mhz but maybe it just wasn't meant to be idk.......im going to focus on getting it stable a 4200mhz in the mean time. 21x instead of 22x.........as much as i hate to do so

What tests are you failing exactly? You have an awfully low Vcore in that screenshot @ 4.4Ghz..Don't be afraid to feed these chips voltage, anything up to 1.45V is fine, although I doubt you would need more then 1.4V to reach 4.5Ghz..Many people will feed them 1.5V+, although I wouldn't recommend that on air, as your one core is already pushing close to 80C..The "Westmere" cores in these CPU's are built on the same process as Sandy Bridge, and we all know how much SB loves Voltage..There are guys that have been running extreme 5Ghz+ O/C's on the 2500/2600K's since launch day, what 3~3.5 years ago?

Are you running 3 X4GB of ram? Or 6x2GB? If it is the latter, your IMC likely needs more voltage, even running the ram @ 1600Mhz.

I heard that IBT is really no good for stability testing, even prime95 apparently. there is some other testing suite that will test every part of the CPU instead of just the narrow band of functions that IBT and P95 test, I forget what software it is though.

Someone recommended the stress test built in to SiSoft Sandra...

What? IBT is fine, as long as you test it properly. You need to run it on "Xtreme Mode" with all available ram. The best way to test with it is to launch the program, and then open task manager and lock the core affinity to the physical cores, in the OP's case Cores 0,1,2,8,9, and 10.

There is nothing wrong with using Prime95 either. It uses all available threads and uses the maximum amount of power of all the stres tests, thus producing the most amount of heat which is key to determining if your O/C is stable.

 

[primetime]

What? IBT is fine, as long as you test it properly. You need to run it on "Xtreme Mode" with all available ram. The best way to test with it is to launch the program, and then open task manager and lock the core affinity to the physical cores, in the OP's case Cores 0,1,2,8,9, and 10.

then open task manager and lock the core affinity to the physical cores, in the OP's case Cores 0,1,2,8,9, and 10.

i never heard of doing this before....what does this achieve?

was a little warmer than usual today and without custom water cooling i do the next best thing:

[/URL][/IMG]

i have passed multiple burn test of hyper pi, sanda, pime95, occt (linpack and the other) but i fail this one about 1 out of 5 times.....pc just turned off like the power supply gave up or something during the very last time i tried this test ( i remember malwabytes asking to do a scan..... i figured wth not?...it was like flipping a switch...im open to ideas as its 100% stable in everything else.....i probably logged over 300 hours on bf4 and never a crash some im not to worried about it. this time it looks like it will pass no problem

[/URL][/IMG]

[/URL][/IMG]


well

 

[Araxie]

then open task manager and lock the core affinity to the physical cores, in the OP's case Cores 0,1,2,8,9, and 10.

i never heard of doing this before....what does this achieve?

was a little warmer than usual today and without custom water cooling i do the next best thing:

it make IBT run without hyperthreading, processing only the true cores, making it more stressful and even making the processor to have better performance. it was discovered long time ago with i5 scoring better speed in IBT due to the way IBT process each core.

 

[primetime]

it make IBT run without hyperthreading, processing only the true cores, making it more stressful and even making the processor to have better performance. it was discovered long time ago with i5 scoring better speed in IBT due to the way IBT process each core.

well ill be damed...i never ever would have knownShit so know i have to do the hole dam test over again because i didn't do it correct? that will have to wait for sleep time...I be honest with yea....what maybe 1% of people that even use IBT even know about this? Or i been living under a rock? Shit...and also you saying my system might not be stable......thanks for the info anyway

 

[primetime]

What tests are you failing exactly? You have an awfully low Vcore in that screenshot @ 4.4Ghz..Don't be afraid to feed these chips voltage, anything up to 1.45V is fine, although I doubt you would need more then 1.4V to reach 4.5Ghz..Many people will feed them 1.5V+, although I wouldn't recommend that on air, as your one core is already pushing close to 80C..The "Westmere" cores in these CPU's are built on the same process as Sandy Bridge, and we all know how much SB loves Voltage..There are guys that have been running extreme 5Ghz+ O/C's on the 2500/2600K's since launch day, what 3~3.5 years ago?

Are you running 3 X4GB of ram? Or 6x2GB? If it is the latter, your IMC likely needs more voltage, even running the ram @ 1600Mhz.




What? IBT is fine, as long as you test it properly. You need to run it on "Xtreme Mode" with all available ram. The best way to test with it is to launch the program, and then open task manager and lock the core affinity to the physical cores, in the OP's case Cores 0,1,2,8,9, and 10.

There is nothing wrong with using Prime95 either. It uses all available threads and uses the maximum amount of power of all the stres tests, thus producing the most amount of heat which is key to determining if your O/C is stable.

i discuss the voltage are another time....but regarding ibt...trying to use your settings it only seems to be using 36% now versus 100% before...what am i doing wrong? This cant be right, what im i misunderstanding? should i disable hyper threading in bios or something?

 

[Araxie]

i discuss the voltage are another time....but regarding ibt...trying to use your settings it only seems to be using 36% now versus 100% before...what am i doing wrong? This cant be right, what im i misunderstanding? should i disable hyper threading in bios or something?

just disable half of available processors. you probably have in the set affinity setup:

CPU 0 -
CPU 1 -
CPU 2 -
CPU 3 -
CPU 4 -
CPU 5 -
CPU 6 -
CPU 7 -
CPU 8 -
CPU 9 -
CPU 10 -
CPU 11 -
CPU 12.

you have to uncheck 1-3-5-7-9-11. that just to make it easy to disable hyperthreading anytime fast without reboot. in any case disable hyperthreading in Bios its the same.

 

[primetime]

just disable half of available processors. you probably have in the set affinity setup:

CPU 0 -
CPU 1 -
CPU 2 -
CPU 3 -
CPU 4 -
CPU 5 -
CPU 6 -
CPU 7 -
CPU 8 -
CPU 9 -
CPU 10 -
CPU 11 -
CPU 12.

you have to uncheck 1-3-5-7-9-11. that just to make it easy to disable hyperthreading anytime fast without reboot. in any case disable hyperthreading in Bios its the same.

seriously this cant be right lol im pushing 22% how is that stressing my cpu?

[/URL][/IMG]


edit...im i doing the wrong process? should it be line pack possibly? this is what i did according to both of your instructions....i apologize if i seem a little "drunk as fuck"

[/URL][/IMG]

 

[ccityinstaller]

seriously this cant be right lol im pushing 22% how is that stressing my cpu?

[/URL][/IMG]

Don't worry about seeing the CPU usage. If you properly disable the hyperthreaded cores, you should sit right @ 50% usage..The important thing is that you watch the GFlops output in IBT..They should be higher then they were in your screenshot above..

It appears that you only have 3 cores active..Notice that 2 of them are running @ 100%, and the third is loaded heavily as well..You may need to play around with which cores to turn off since a Xeon is different then the native 4 core dies with HT..with those, you just do like Axarie said and turn off the odd number cores..

In your case, based on your screenshot from before, it appears that cores 0,1,2 and 8,9,10 are the physical cores in your CPU..Try locking IBT to those cores, and monitor the GFlop output..it should be higher then those results in your first screenshot.

It seems like you are stable, if you can pass everything you have and been able to game without issues as well. The issue where the system was doing a hard power off when loaded is most likely due to the Vcore being too low..

I had that issue when fine tuning my 3770K on my MSI Z77 MPower..Everything would be stable for hours and hours, and then I would get a random hard reboot. I had to boost my Vcore and then give my IMC a tiny bit of juice as well.

 

[Deimos]

Primetime how in the hell are your temps so low? according to my temp monitor my X5670 idles at 31C load as high as 85C and I have a pretty hardcore watercooling setup, what gives?

I'm starting to think that maybe the method I use to apply thermal paste is not ideal, I'm going to have another crack at it, any tips from your obviously superior method?

 

[Deimos]

Sorted. Looks like I put too much AS5 on there, I was using the "mound" method.

Switch to some generic looking prolimatech compound I had lying around and switched back to the "thin spread" method and my temps dropped by 15C

Further tweaking might be warranted, I'm not sure the water block is applying good pressure on the socket and my application of the compound was hasty, I'll have another fiddle later on tonight and see what happens.

I can definitely see room for improvement, just noticed that half my cores have now gone 5C higher than before I changed the compound.

 

[Deimos]

Well f*** me, I didn't install the WB properly, the bracket wasn't quite stiing right and it popped in to place leaving the WB loose. Thank f*** for throttling, temps instantly went to 95 and throttling kicked in.

Also after I pulled the WB off again to fix it I noitced that the middle of the WB is not making contact with the heat spreader at all, probably because I lapped it a few years back, making it flat. Looks like I might have to lap my CPU. Fun times ahead.

 

[schizrade]

My L5639 on a Sabertooth is an absolute beast for a gaming machine. If x99 is less problematic than X79 I might upgrade, but this is still so relevant I might just sit on it a while longer.

 

[primetime]

Primetime how in the hell are your temps so low? according to my temp monitor my X5670 idles at 31C load as high as 85C and I have a pretty hardcore watercooling setup, what gives?

I'm starting to think that maybe the method I use to apply thermal paste is not ideal, I'm going to have another crack at it, any tips from your obviously superior method?

im using a very high grade thermal grease...plus in in the pic i m kinda cheating with an small fan blowing fresh air directly at the heat sinks....and ive had a good bit of practice applying grease the correct way...most people are giving incorrect instructions on how to install the grease.....you just want a very thin even layer on both surpasses...to much or too little will not give good results...not to mention it has to mounted in near perfect flat contact...otherwise it burns up....this hole putting a dot hear and there is retarded...its has to be spread perfectly evenly to do its job

 

[primetime]

What tests are you failing exactly? You have an awfully low Vcore in that screenshot @ 4.4Ghz..Don't be afraid to feed these chips voltage, anything up to 1.45V is fine, although I doubt you would need more then 1.4V to reach 4.5Ghz..Many people will feed them 1.5V+, although I wouldn't recommend that on air, as your one core is already pushing close to 80C..The "Westmere" cores in these CPU's are built on the same process as Sandy Bridge, and we all know how much SB loves Voltage..There are guys that have been running extreme 5Ghz+ O/C's on the 2500/2600K's since launch day, what 3~3.5 years ago?

Are you running 3 X4GB of ram? Or 6x2GB? If it is the latter, your IMC likely needs more voltage, even running the ram @ 1600Mhz.




What? IBT is fine, as long as you test it properly. You need to run it on "Xtreme Mode" with all available ram. The best way to test with it is to launch the program, and then open task manager and lock the core affinity to the physical cores, in the OP's case Cores 0,1,2,8,9, and 10.

There is nothing wrong with using Prime95 either. It uses all available threads and uses the maximum amount of power of all the stres tests, thus producing the most amount of heat which is key to determining if your O/C is stable.

is this kinda like the way you described? i f i set it to max the task manager starts going crazy and wont work right.....never in my life see it do that before..... i went and tried 0-5 to see how that works....at least its at 50% usage

[/URL][/IMG]

like you said i may have to find which are the actual cores and which are just hyper cores? very interesting and crazy i never heard of this method before

 

[USF-Nealio]

Subscribed. I'll need to post my rig.

I have a L5639 in an Asus Rampage III Extreme, and even though it's overclocked slightly, none of my games have gotten close to maxing out the CPU. I'll need to upgrade my graphics beyond a single HD 7950 before it'll break a sweat.

 

[Deimos]

im using a very high grade thermal grease...plus in in the pic i m kinda cheating with an small fan blowing fresh air directly at the heat sinks....and ive had a good bit of practice applying grease the correct way...most people are giving incorrect instructions on how to install the grease.....you just want a very thin even layer on both surpasses...to much or too little will not give good results...not to mention it has to mounted in near perfect flat contact...otherwise it burns up....this hole putting a dot hear and there is retarded...its has to be spread perfectly evenly to do its job

Which thermal grease do you recommend? I had another crack late last night, a thin layer on the CPU and an even thinner layer on the WB but I know that the middle of the WB is not making any contact with the centre of the CPU because I lapped it (I lapped a Q6600 back in the day with this same WB). I'm going to lap my X5670 since this is going to be the last CPU in this system. I'll do it sometime this week and report back with results.

I'm still fiddling, my new settings are 24x183 BCLK, I'm experimenting with ram overclocking, the G.Skill ram I have seems to be holding up quite well and there was a nice 4GB/s improvement in memory bandwidth.
I'm picking up a new PSU today, wonder if it will make any difference to overclocking, the PSU I'm currently running is still out of spec on the 3.3V, just not as bad as the last one I had. I'm going with a Seasonic Platinum 1000W.

 

[Deimos]

Subscribed. I'll need to post my rig.

I have a L5639 in an Asus Rampage III Extreme, and even though it's overclocked slightly, none of my games have gotten close to maxing out the CPU. I'll need to upgrade my graphics beyond a single HD 7950 before it'll break a sweat.

It has been a long time since video cards were CPU bound, you'd have to go back to games like Half-Life 2 or possibly even Quake 3 Arena.

I'm playing Watch Dogs at the moment and my CPU doesn't even run at its highest possible clock half the time I'm playing. It is heavily threaded but again, it's only using about 40% of my cores.

The only game I play that it CPU bound at the moment is Shift 2, but that is only because the video card is not at 100% and Shift 2 isn't threaded so I see one core at 100% all the time.

 

[ND40oz]

This was the limit of my P6T6 WS Revo, the L5609 probably could have gone higher if I hadn't hit the bus limit.

http://hwbot.org/submission/1064164_nd40oz_cpu_frequency_xeon_l5609_2866.7_mhz

 

[schizrade]

It has been a long time since video cards were CPU bound, you'd have to go back to games like Half-Life 2 or possibly even Quake 3 Arena.

I'm playing Watch Dogs at the moment and my CPU doesn't even run at its highest possible clock half the time I'm playing. It is heavily threaded but again, it's only using about 40% of my cores.

The only game I play that it CPU bound at the moment is Shift 2, but that is only because the video card is not at 100% and Shift 2 isn't threaded so I see one core at 100% all the time.

The only game that pushes my CPU that I play normally is CIV V especially when running huge maps late in game.

 

[USF-Nealio]

It has been a long time since video cards were CPU bound, you'd have to go back to games like Half-Life 2 or possibly even Quake 3 Arena.

I'm playing Watch Dogs at the moment and my CPU doesn't even run at its highest possible clock half the time I'm playing. It is heavily threaded but again, it's only using about 40% of my cores.

The only game I play that it CPU bound at the moment is Shift 2, but that is only because the video card is not at 100% and Shift 2 isn't threaded so I see one core at 100% all the time.

With the strange exceptions like Arma (and to a lesser extent BF4), you are absolutely right. I was able to play 98% of games at stock clocks, and was only overclocking to overcome the age of my CPU on those specific titles.

I upgraded from an i5-750 to the L5639. When I played BF4 on the i5-750 @4.2GHz, I was pegged at 100% CPU on all 4 cores while my GPU utilization fluctuated around 70%. With the stock L5639, I'm only seeing 33% CPU utilization and 100% GPU.

 

[primetime]

Which thermal grease do you recommend? I had another crack late last night, a thin layer on the CPU and an even thinner layer on the WB but I know that the middle of the WB is not making any contact with the centre of the CPU because I lapped it (I lapped a Q6600 back in the day with this same WB). I'm going to lap my X5670 since this is going to be the last CPU in this system. I'll do it sometime this week and report back with results.

I'm still fiddling, my new settings are 24x183 BCLK, I'm experimenting with ram overclocking, the G.Skill ram I have seems to be holding up quite well and there was a nice 4GB/s improvement in memory bandwidth.
I'm picking up a new PSU today, wonder if it will make any difference to overclocking, the PSU I'm currently running is still out of spec on the 3.3V, just not as bad as the last one I had. I'm going with a Seasonic Platinum 1000W.

[/URL][/IMG]

[/URL][/IMG]

[/URL][/IMG]

ill just stick with normal way for Intel burn test....seems to to the job quite well.....still looking stable at 4400mhz with max temps at 75C......lol that works fine for me...especially using 6 yeal old air cooler...but like i said before i was cheating a little with a fan blowing air toward my case with the side off......not like any game will ever push it that far anyway

 

[primetime]

Subscribed. I'll need to post my rig.

I have a L5639 in an Asus Rampage III Extreme, and even though it's overclocked slightly, none of my games have gotten close to maxing out the CPU. I'll need to upgrade my graphics beyond a single HD 7950 before it'll break a sweat.

welcome to the club...more the better....lets see how far we can all push these golden cpus.....even the newer ones cant keep up many times

 

[primetime]

What tests are you failing exactly? You have an awfully low Vcore in that screenshot @ 4.4Ghz..Don't be afraid to feed these chips voltage, anything up to 1.45V is fine, although I doubt you would need more then 1.4V to reach 4.5Ghz..Many people will feed them 1.5V+, although I wouldn't recommend that on air, as your one core is already pushing close to 80C..The "Westmere" cores in these CPU's are built on the same process as Sandy Bridge, and we all know how much SB loves Voltage..There are guys that have been running extreme 5Ghz+ O/C's on the 2500/2600K's since launch day, what 3~3.5 years ago?

Are you running 3 X4GB of ram? Or 6x2GB? If it is the latter, your IMC likely needs more voltage, even running the ram @ 1600Mhz.




What? IBT is fine, as long as you test it properly. You need to run it on "Xtreme Mode" with all available ram. The best way to test with it is to launch the program, and then open task manager and lock the core affinity to the physical cores, in the OP's case Cores 0,1,2,8,9, and 10.

There is nothing wrong with using Prime95 either. It uses all available threads and uses the maximum amount of power of all the stres tests, thus producing the most amount of heat which is key to determining if your O/C is stable.

Actually this info is out dated.....i wish it wasn't...but 32nm westmere/gulftown can be damaged going above 1.35 vcore ......yea the older ones could take a lot higher voltage with no damage at all with good temps...but things are a bit different for the 32nm cpus....they cant Handel the same juice the older ones could without damage...or at least thats what i been told/read lately.....believe me i would love to throw 1.4 volts at it and go for 4600mhz

edit: oh and regarding the ram it must be the 2gb sticks.....6 total in tipple channel for 12gb total

 

[primetime]

With the strange exceptions like Arma (and to a lesser extent BF4), you are absolutely right. I was able to play 98% of games at stock clocks, and was only overclocking to overcome the age of my CPU on those specific titles.

I upgraded from an i5-750 to the L5639. When I played BF4 on the i5-750 @4.2GHz, I was pegged at 100% CPU on all 4 cores while my GPU utilization fluctuated around 70%. With the stock L5639, I'm only seeing 33% CPU utilization and 100% GPU.

[/URL][/IMG

Same here lol......people always whining how cpu intensive it is.....mine barley breaks a sweat at around 30% usage......temps for cpu were 33c while playing 64 man server large map with what seems like a lot of weather physics
I think our cpus from 5 years ago still smoke newest amd cpus to be got....there cards are great....i have everything maxed out and mostly says around 75fps

Do they even make a game that would push are xeon cpus...none that i have found yet....from looking at the osd it really show i could use an r9 290x.....even though its not really ended

 

[seanmcd]

I have access to many of these old CPU's - but what's the ideal board that's not too hard to acquire on ebay for these? Asus P6T?

 

[Deimos]

There is a few on there, I picked up a P6X58D Premium not too long ago, apparently the P6X58D-e variant is better. The Rampage 3, P6T.

 

[CrimsonKnight13]

Subscribed. I'll need to post my rig.

I have a L5639 in an Asus Rampage III Extreme, and even though it's overclocked slightly, none of my games have gotten close to maxing out the CPU. I'll need to upgrade my graphics beyond a single HD 7950 before it'll break a sweat.

I'll post mine as well (maybe after I have a R9 290 installed). I have my L5639 OC'd to 3.6. Past that I was having issues with RAM clock stability. I'd still love to OC it to 4GHz at least but I'm extra cautious about doing that. No golden parachutes here when I make a mistake with the technology I own.

No games have taxed it much, so I'm very happy with having it.

By the way... does anyone have recommended OC settings for the ASUS Rampage III Formula with the L5639? I'd like to compare any tips with what I've got running.

 

[Deimos]

The ram overclocking was a bust, my sound card suddenly didn't like it after I swapped out the PSU for the Seasonic, back to 22x200

It was hard locking my PC, couldn't even boot in to windows but as soon as I took out the sound card it was fine again.

 

[Deimos]

I'll post mine as well (maybe after I have a R9 290 installed). I have my L5639 OC'd to 3.6. Past that I was having issues with RAM clock stability. I'd still love to OC it to 4GHz at least but I'm extra cautious about doing that. No golden parachutes here when I make a mistake with the technology I own.

No games have taxed it much, so I'm very happy with having it.

By the way... does anyone have recommended OC settings for the ASUS Rampage III Formula with the L5639? I'd like to compare any tips with what I've got running.

What are your current settings? and what is the max multiplier?

 

[primetime]

i failed again.......4400mhz is so close to being 100% stable butt fails 1 out of 5 stress tests ......doping the mult to 21 fror 4200mhz....and well see if it things really sorted out the issues
i will say this its never bsod my system or crashed for that matter...but its got to be 100% stable...maybe if its was winter it would be ok......who knows?
if it passes everything i can throw at it at 4200mhz.....i still be happy.....could just be that one bios setting im missing in bios fro perfection .......idk thats what this thread is for
those who want to compete: show screen shots of intel burn test passing, hyper pi, sanrda, and what ever else proves s beyond any doubt it good to go .......those with rampage will likely win every time

ill do more real test tonight...for now its bf4 time!!!!!! i love the fact that we have 6 yea old systems that beat the newest tech out