1366 x58 Xeon Enthusiast overclocks club

With enough RAM you can comfortably turn off your page file and still multitask like a mofo. You could set up a RAM disk if you wish. Other than that, it just depends on your workload... being able to keep a VM or some other stuff running while you take a gaming break is nice.
 
FlawleZ said:
This. The X56xx have double the QPI bandwidth for the ability to run dual processors whereas the W series can only run 1.
Click to expand...

Hmm, I'm not sure if that has anything to do with the QPI/Rated ram speed. The w3680 is rated at 6.4GT/s and officially supports 1333mhz ram but doesn't support a dual CPU config.

Not that any of that means anything. When overclocking things get pushed much higher anyway. For example my QPI is running at 7.2 GT/s and my ram is at 2000mhz.

But I do wonder how the w3670 will fare at those kind of speeds. Luckily it has access to two more multi's so the BCLK can drop quite a bit to get the same overclock.
 
Bit of an x-post from OCN, but seems there's more combined knowledge about these xeons on 1366 boards here. I've got an Asus P6T, and 3x4gb 1600mhz corsair ram modules. Just upgraded from a 920 to a x5670. After setting in the new chip, seemed to be able to boot, but would only read 8gb of memory. Updated to 1408 BIOS, reset BIOS, pulled battery, reseated CPU, rotated modules (1x4gb stick, tried each stick individually) but system still only reads my ram as 8gb 1333mhz. Setting the ram to 1600 manually results in a "overclocking failed" after the POST runs at 1333 again. Trying manual DRAM voltage 1.5, same thing. Not super worried about the speed just yet, would rather get the full ram capacity running, but thought it might be useful info. Seems like an awesome CPU, runs stupid cool on the new lithography despite the extra cores and stock frequency, can't wait to see what it does overclocking.
 
angrysasquatch said:
Bit of an x-post from OCN, but seems there's more combined knowledge about these xeons on 1366 boards here. I've got an Asus P6T, and 3x4gb 1600mhz corsair ram modules. Just upgraded from a 920 to a x5670. After setting in the new chip, seemed to be able to boot, but would only read 8gb of memory. Updated to 1408 BIOS, reset BIOS, pulled battery, reseated CPU, rotated modules (1x4gb stick, tried each stick individually) but system still only reads my ram as 8gb 1333mhz. Setting the ram to 1600 manually results in a "overclocking failed" after the POST runs at 1333 again. Trying manual DRAM voltage 1.5, same thing. Not super worried about the speed just yet, would rather get the full ram capacity running, but thought it might be useful info. Seems like an awesome CPU, runs stupid cool on the new lithography despite the extra cores and stock frequency, can't wait to see what it does overclocking.
Click to expand...

Try setting all your timings manually and work from there.

To run 1600mhz in my board the BCLK has to be 160mhz or higher, which is why I only run my 2400mhz modules at 2000mhz (so that I don't stress the board too much). If I pump up my BCLK to 220mhz I can run them at 2200mhz no problem.

I don't know if it's a limitation with the platform, but I have the same issue on my ASRock x58 Extreme.
 
I'm also going to cross post here from OCN but only to maximize views and opinions and hopefully help someone else in the future if they run into this. So here it goes:

I have an interesting case for you guys to hear and tell me what you think the problem is.

Several weeks back a friend hit me up and wanted my MSI Big Bang X58 board in trade for his Asus Sabertooh X58. At the time I hadn't yet picked up my X5675 but the board had been chugging along nicely with an i7 950. His Sabertooth had been working fine with a 980X. So, we swap boards and use our respective 6 core CPUs perfectly for a few weeks until a couple days ago he says he arrives at his PC to see it only a black screen and blinking cursor. Calls me up and we go through all the troubleshooting. I go over to his house to continue troubleshooting and what we found was:

The board would always boot to the BIOS without issue every time. After BIOS splash screen and the Marvell SATA controllers load, it would hang with the same blinking cursor. Initially we suspected OS or drive failure, but it was more than that. We eventually noticed that for a split second the board would flash "Checking NVRAM" before it would hang. We pulled every drive, every cable, every stick of RAM, etc. You name it we did it. Every drive and peripheral would detect perfectly in the BIOS, USB, SATA, etc. Nothing would boot or load beyond that extremely brief "Checking NVRAM" message.

Our last piece to change out was the CPU. We put an i7 975 in the board and boom, it booted to Windows beautifully. So we went back to the 980X, and tried disabling HT and cores. What we found was the system would boot and function perfectly if the 980X was forced to 1 core, 2 cores, or 3 cores. 4, 5, and 6 cores would cause the system to hang.

Here's where you come in: I personally think its the 980X and something failed, causing only 3 cores to function properly. He maintains he thinks the 980X is fine and that the Big Bang has some odd compatibility issue with the 980X alone. We can't seem to find anyone else online mentioning a compatibility issue with this CPU and board before. I'm heading over this evening with the Sabertooth and X5675 to test to find out for sure.

What do you guys think? Anyone ever heard of a 980X losing half its cores? What about compatibility issues?
 
angrysasquatch said:
After setting in the new chip, seemed to be able to boot, but would only read 8gb of memory. Updated to 1408 BIOS, reset BIOS, pulled battery, reseated CPU, rotated modules (1x4gb stick, tried each stick individually) but system still only reads my ram as 8gb 1333mhz.
Click to expand...

Set everything to auto, reseat the ram sticks, reseat your CPU again and make sure there are no bent pins and CPU is not overtightened by your cooler.

I had same shit, problem was one of the sticks was not seated correctly :eek: also have or had 3 bent pins but fortunately those are some unused pins for VCC which I think I managed to straightened out with a needle :D

Good luck!
 
Quick Google search gave some hits with people running the MSI X58 Big Bang with a 980X without problems. I would double check that you have the latest BIOS (or a version that officially supports the 980X), do a BIOS reset, and downclock the memory. I guess it's possible that the CPU was designed in such a way that if you disabled a damaged core then it would be totally dissociated from all the other shared buses, caches, and power rails and everything operates just fine. I guess it gave Intel a way to waterfall a nice high end CPU into something else. Maybe run it with 3 cores, stress test it, and see if anything else falls out.



On a completed unrelated note I just ordered a Xeon x5675 to replace my old 920 that I couldn't get past 3.8GHz no matter what I tried. Hopefully it overclocks nicely and I'll see some gains in CPU limited games.
 
Killdozer said:
Quick Google search gave some hits with people running the MSI X58 Big Bang with a 980X without problems. I would double check that you have the latest BIOS (or a version that officially supports the 980X), do a BIOS reset, and downclock the memory. I guess it's possible that the CPU was designed in such a way that if you disabled a damaged core then it would be totally dissociated from all the other shared buses, caches, and power rails and everything operates just fine. I guess it gave Intel a way to waterfall a nice high end CPU into something else. Maybe run it with 3 cores, stress test it, and see if anything else falls out.



On a completed unrelated note I just ordered a Xeon x5675 to replace my old 920 that I couldn't get past 3.8GHz no matter what I tried. Hopefully it overclocks nicely and I'll see some gains in CPU limited games.
Click to expand...

Thanks for the reply. Yes, I found a few here and there mentioning the 980X and Big Bang working out. The BIOS is only 1.7 which is the latest from MSI. The MSI board booted seemed fine with the i7 975 isntalled. He's also had it up and running for a day or so on 3/6 cores on the 980X with HT enabled. I learned this afternoon he even stressed those 3 cores on the 980X today at 4.6Ghz without any problems. This system runs all higher end components from a few years back as you can imagine. Corsair XMS 1600Mhz, SLI GTX 780's, Asus Xonar, OCZ Revodrive2, etc.

And you'll love the X5675. Mine does 4Ghz effortlessly at 1.25V and scales fairly up to 4.6Ghz. 4.8Ghz boots but I don't have it stable yet and may not with just an H100i GTX. 5Ghz boots and I can complete single threaded CB and SuperPi but anything more isn't stable.
 
Got it working, thanks guys. Didn't pay much attention to it last time I wiped the CMOS, but after POST it wanted me to set things up, but the clock was still on time. Must've wiped it properly this time after pulling the CPU and stuff, and the time needed to be set, all working now and at 4ghz, 1.25v... for now :D
 
Just update if anyone else runs into the issue I described above, the 980X was in fact bad. We installed it in the Sabertooth and the symptoms were exactly the same. It seems as if something can fail on the 6 core chips causing only half to work or fail. We ended up ordering an X5675 last night to replace it.
 
FlawleZ said:
Just update if anyone else runs into the issue I described above, the 980X was in fact bad. We installed it in the Sabertooth and the symptoms were exactly the same. It seems as if something can fail on the 6 core chips causing only half to work or fail. We ended up ordering an X5675 last night to replace it.
Click to expand...

Thanks for the update!
 
FlawleZ said:
What are you doing that exceeds 12GB? Certainly not gaming...
Click to expand...

I sit at 8-10GB used most days. Just from tabs open in Chrome, music playing, having Lightroom open and Chrashplan backing up 9TBs of stuff. Crashplan is a bit of a memory hog with very large filesets like I have. Using 4GB for 700,000 files. Once I start editing photos in Lightroom and maybe get a image opened in Photoshop I'm certainly hitting up against the 12GB.

I installed a SSD in February which helped smooth out everything. Felt like I had upgraded the ram. Performance still slows down if I get a big panorama opened, or a file in Photoshop with many layers, but there is less stuttering than before.
 
MN Scout said:
I sit at 8-10GB used most days. Just from tabs open in Chrome, music playing, having Lightroom open and Chrashplan backing up 9TBs of stuff. Crashplan is a bit of a memory hog with very large filesets like I have. Using 4GB for 700,000 files. Once I start editing photos in Lightroom and maybe get a image opened in Photoshop I'm certainly hitting up against the 12GB.

I installed a SSD in February which helped smooth out everything. Felt like I had upgraded the ram. Performance still slows down if I get a big panorama opened, or a file in Photoshop with many layers, but there is less stuttering than before.
Click to expand...

Yep same here. Doing web development work, I never realized how much ram I would use during a typical work day until I went to 24gb. I typically sit between 8-16gb used depending on my work load (Photoshop, VM's (for testing different platforms), Chrome, Firefox, IE).
 
I just revived my old 1366 system with a x5675, and my 6 sticks of 2GB RAM I had in the old system no longer seem to run at 1600MHz at the designated 1.65v (they ran fine with an i7 920) -- so I began looking into replacing all 6 banks with just 3 DIMMs of 8 GB as an upgrade, and to not run the IMC as hard. I was concerned that perhaps the motherboard wouldn't support the full 8 GB per channel, but after a lot of Googling and lurking this entire thread, my concerns were unwarranted. HOWEVER, through my research I ran into an interesting thing about some X58 motherboards (This is a Gigabyte GA-X58-UD5) and the older i7 memory controllers -- apparently you can configure tri-channel memory utilizing 4 DIMM?

From the manual:
3 Channel--
1. 3 Channel mode cannot be enabled if only one or two DDR3 memory modules are installed.
2. When enabling 3 Channel mode with three, four or six modules, it is recommended that memory
of the same capacity, brand, speed, and chips be used.
When enabling 3 Channel mode with three memory modules, be sure to install them in the
DDR3_1, DDR3_3 and DDR3_5 sockets.
When enabling 3 Channel mode with four memory modules, be sure to install them in the
DDR3_1, DDR3_2, DDR3_3 and DDR3_5 sockets.
Click to expand...

So with interleaved memory (tri-channel, four DIMM) what kind of measured performance impact does this have on the platform? I have 3x Crucial Tactical CL8 1.35v sticks headed my way, and I was wondering if I could gain anything by simply ordering a 4th stick? (Aside from the obvious, going from 24 to 32 GB RAM)
 
MN Scout said:
I sit at 8-10GB used most days. Just from tabs open in Chrome, music playing, having Lightroom open and Chrashplan backing up 9TBs of stuff. Crashplan is a bit of a memory hog with very large filesets like I have. Using 4GB for 700,000 files. Once I start editing photos in Lightroom and maybe get a image opened in Photoshop I'm certainly hitting up against the 12GB.

I installed a SSD in February which helped smooth out everything. Felt like I had upgraded the ram. Performance still slows down if I get a big panorama opened, or a file in Photoshop with many layers, but there is less stuttering than before.
Click to expand...

Ok so very heavy multitasking. I can see that. You're doing the work of what used to take 2-3 PC's on just 1. :p On that note, I would suggest a 3X8GB setup for your system.
 
Well, got my w3670. It's a bit disappointing, it overclocks about on par with the x5670 in my asrock board but is much hotter.

4.2ghz 175x24 needed 1.325v. It hit the low 80's in small ffts pretty quick so I stopped the test, this was also with the QPI at 1.2v and everything as low as possible other than the clockspeed.
I messed around with it for about an hour until I gave up and went back to my x5670.

Thermal paste cover/thickness looked good when I removed the heatsink. Idle temps were quite high as well, hitting the mid 40's while my x5670 at 4.2 idles around 30c.
Even running Cinebench resulted in upper 70's. My x5670 at 4.2 maxes at 55c in Cinebench right now.

Oh well, it was worth a try for the price. I'll probably sell it as I can make a few bucks.
 
tbob22 said:
Well, got my w3670. It's a bit disappointing, it overclocks about on par with the x5670 in my asrock board but is much hotter.

4.2ghz 175x24 needed 1.325v. It hit the low 80's in small ffts pretty quick so I stopped the test, this was also with the QPI at 1.2v and everything as low as possible other than the clockspeed.
I messed around with it for about an hour until I gave up and went back to my x5670.

Thermal paste cover/thickness looked good when I removed the heatsink. Idle temps were quite high as well, hitting the mid 40's while my x5670 at 4.2 idles around 30c.
Even running Cinebench resulted in upper 70's. My x5670 at 4.2 maxes at 55c in Cinebench right now.

Oh well, it was worth a try for the price. I'll probably sell it as I can make a few bucks.
Click to expand...

Interesting there was that big of a difference. I guess its just the 95W vs 130W TDP
 
I guess I'm a glutton for punishment. I'm going to try and get to 48GB again.

Not finding any listings 3x8gb or 6x8gb kits. Checked ebay and the prices for used non-ecc and non FB-dimm sticks are the same as what I found on Amazon. Appears the price for new DDR3 has been falling since January. Best deal I have found is for a Crucial Ballistix Sport 16GB Kit (8GBx2) at $78. I bought 3 of the 2x8GB kits.

Should be here Monday. Will probably take all night to get it going. I'll report back.

Anyone know if Amazon takes back ram OK if this test doesn't work, or will I need to sell here on the forums? Hopefully at least 24GB will work and I can sell the rest here.

Edit: WTF Chrome? Why are you using 200MBs of ram for each page? Currently have 500MB "free" and I'm not evening doing any photo editing.
 
Last edited:
As an Amazon Associate, HardForum may earn from qualifying purchases.
FlawleZ said:
Interesting there was that big of a difference. I guess its just the 95W vs 130W TDP
Click to expand...

Could be.. I wasn't really convinced that it would be that big of difference. The first x5670 I got was similar but overclocked far worse. That thing got up in the mid 80's at 4ghz and needed 1.35v.. Of course I immediately sent it back and gladly paid the restocking fee.

The second one was the one I'm using now and is the best I've had yet out of the six I've tried, it's the only one that is stable at 4.4ghz near 1.325v.. Most of the others I've tried would need at least 1.35v or maybe even 1.4v.
 
Hmm, the problem with not recognizing all my memory is back. When it's doing that, it also only boots half the time. For instance, if I save something in the BIOS and it restarts, it won't actually boot, I'll just have a blank screen. Same with shutting or restarting from windows, needs to be started once to an unresponsive state, forced to shut down and the next boot everything is normal, except the ram thing. CMOS reset and pulling the CPU from the socket did not work this time. Might there be something wrong with the CPU itself? Would like to know for sure before the seller's warranty runs out.
 
angrysasquatch said:
Hmm, the problem with not recognizing all my memory is back. When it's doing that, it also only boots half the time. For instance, if I save something in the BIOS and it restarts, it won't actually boot, I'll just have a blank screen. Same with shutting or restarting from windows, needs to be started once to an unresponsive state, forced to shut down and the next boot everything is normal, except the ram thing. CMOS reset and pulling the CPU from the socket did not work this time. Might there be something wrong with the CPU itself? Would like to know for sure before the seller's warranty runs out.
Click to expand...

Could be the memory controller dying. I'm assuming you've tested without overclocking anything? Have another CPU to test with?
 
Vlad_13 said:
That would be an extremely rare thing to occur... My bet would be on a physical issue.
Click to expand...

Depends on what kind of overclock. I've heard of more than a few cases of running high overclocks with high QPI voltages and sticks disappearing. But yeah, something like that is more likely to happen with either bent pins or the CPU not making very good contact.

I've actually had it happen when I was running my timings tighter than some of my sticks could handle.
 
Last edited:
Yeah, took off the overclock. It's been running the 920 flawlessly at stock for years now, I'll drop it back in to check. Pins and CPU base look fine, I look them over carefully for defects every time I pull the CPU. New CPU has only been clocked to 3.99, 1.25 vcore, 1.25 qpi, slightly underclocked ram at stock timings and 1.5 (stock) volts.
 
IF you desire informed help, You should consider putting relevant hardware in your sig. I personally have no idea what hardware your using to even begin to comment
 
Ah, was in the OP, guess it's on the other page now. I'll do it. edit: done, assuming I can't see my own signature. Let me know if it's not showing up for you guys.

12gb of DDR3-1600 with the 920, no issues. Here's the socket:

2d0e1dw.jpg
 
Last edited:
There is some dark grey thermal paste on the bottom left corner... Also there is some crap under a pin between 1st and 2nd column in the top middle, looks like some more of the thermal paste, cant tell tho... Clean em CAREFULLY with some home-made micro q tips (using toothpicks) with rubbing alcohol. If you cant reach em use toothpicks alone without cotton.

Other than that socket looks fine to me, but cant say about that thermal paste tho... It might be conductive.
 
Last edited:
Vlad_13 said:
There is some dark grey thermal paste on the bottom left corner... Also there is some crap under a pin between 1st and 2nd column in the top middle, looks like some more of the thermal paste, cant tell tho... Clean em CAREFULLY with some home-made micro q tips (using toothpicks) with rubbing alcohol. If you cant reach em use toothpicks alone without cotton.

Other than that socket looks fine to me, but cant say about that thermal paste tho... It might be conductive.
Click to expand...

What thermal paste electrically conductive these days? The last one I can remember is Arctic Silver 3.

The stuff in bottom left corner looks like dust buildup to me, I actually have some of that on the bottom of my socket. A can of compressed air may get it out, or it may move it into a worse spot making it more difficult to get out.

Whatever is in-between the first and second rows from the left in the middle area does look like something that could cause issues, I'd use compressed air very carefully and see if it will fly out.
 
Has anyone else tried to get Win10 going yet? Upgrade install from Win7 x64 went fine, but clean install freezes on setup. I've installed Win10 about 10x now, I get further by installing from a DVD than from USB.

Hardware:
Asus P6T Deluxe V2 with newest 1202 bios, set to defaults
Xeon x5670 at default speeds for install
Corsair 16GB ddr3 dual channel, underclocked to x58 default 1333mhz
Crucial M4 128GB ssd with newest 070H firmware
Asus Strix GTX980, changed to DVI cable from displayport just in case
Intel CT gigabit pci-e nic
basic Spawn mouse and Leopold mech keyboard
all other peripherals/drives have been disconnected

I've tried:
- disabling all Virtualization settings in bios (reported to be an issue)
- disabling my onboard Marvell ethernet controllers (reported to be issues with stock drivers)
- changing from Media Creation Tool to previous Windows USB/DVD Download Tool from MS, tried both USB and DVD methods
- tried different mech keyboards in case there's a conflict with a controller

At first, after USB install and reboot I would get a loading screen for 2 seconds and then a black screen forever. Switching to other USB method resulted in the same screen. Reboots would result in the same screen. Installing from a DVD would do the same, but on reboot it would let me choose some setup options, setup for my hardware, reboot, and then freeze when trying to update (not even a login screen yet). Monitoring traffic on my router shows no traffic, despite the lights blinking for a minute. Leaving it for half an hour with still no change in the screen (animation still frozen), forcing a shutdown and reboot, it gets frozen again at another screen that looks like it's trying to do initial login.

I've installed Win7 again to see if something's messed up, went fine. Ran through Memtest86+ and passed. Try to reinstall 10 again and hit the same frozen spots.

I've seen one reference to Win10 not liking x58 and ssd's on a clean install, and installing to hdd worked for them.

This has eaten up my whole weekend. To add insult to injury my internet dropped out today so I had to look stuff up on my phone, lol.
 
Was yours an uneventful install, or painful like mine?

So it seems that the trick on mine was to wait at least 30 mins while frozen, and then force shutdown and boot. Somehow, this time it would get a bit farther in the setup process before freezing. I'm now posting this from Edge, I finally got in!!!
 
RickyJ said:
Was yours an uneventful install, or painful like mine?

So it seems that the trick on mine was to wait at least 30 mins while frozen, and then force shutdown and boot. Somehow, this time it would get a bit farther in the setup process before freezing. I'm now posting this from Edge, I finally got in!!!
Click to expand...

Nice!

I went from 8.1>10, I didn't change any bios settings, left it overclocked. I didn't have any major issues, I had some speed issues with my 390 at first but clearing the drivers out sorted that out. I did the standard in place upgrade, not from USB.
 
RickyJ said:
Was yours an uneventful install, or painful like mine?

So it seems that the trick on mine was to wait at least 30 mins while frozen, and then force shutdown and boot. Somehow, this time it would get a bit farther in the setup process before freezing. I'm now posting this from Edge, I finally got in!!!
Click to expand...

Thats crazy..we have same board, same bios, and i have never had a single windows 10 install issue yet, all the way from 1st build to rtm. Course i never once changed a single bios setting....overclocked all the way thru:D Now i imagine you don't use the same bios settings i do? Maybe thats it?
 
Yeah this has been consistently shit. I first tried with only dropping my bclk to 133, everything else with same settings. In keeping with the trend, 10 won't activate because my key is "blocked". Did reactivating 7 erase my hardware-locked activation key on MS servers? I've tried forcing activation using "slmgr.vbs /ato" from elevated prompt, but just get error code 0xC004C003. Apparently I should wait a day to see if the servers clear up, sooooo fingers crossed?
 
Try doing a long clear c mos sometime. Mine was acting up a couple years ago...tried everything...nothing worked...did a long clear c moss, re did all settings and was good to go ever since. Its like the bios rom gets a glitch is only way i can describe it
 
Yeah I didn't try that, might have helped after removing all of my cards/peripherals. This machine has just been so stable after replacing the failing 920 that I forgot about it, lol. If it doesn't activate tomorrow, I'll put 7 back on and upgrade then do a clean install from a fresh DVD.
 
Woah yeah! Booted right up with the 48GB of Crucial balistix ddr3 1600 ram. Before I installed the ram I reset the bios options for the ram and cpu. Once it was up I loaded my OC profile back in. Currently running Memtest at 4GHz. This ram is faster then my old stuff, so I could probably try for a higher OC - ram is running at 1440 right now.

The bios feels "smoother" now :p.
 
I decided to get a kill a watt to see how power hungry this platform really is. It's not nearly as bad as I was expecting.

Idle: 145w
Lots of USB stuff plugged in, all of the stuff in my sig plus a USB3 card, 5 120mm fans, and 3 140mm fans

Prime95 Small FFTs: 310w
Furmark: 495w
Furmark+Small FFTs (bad idea?): 660w

This is with the 390 at stock.. That is some crazy usage, but I guess it's about what I'd expect.

I'll see what normal load is when I get a chance.. Obviously this is pretty much the highest wattage it will every see.

The accuracy is debatable, but I'm not going to spend $100+. The kill-a-watt was $10. :)
 
You must log in or register to reply here.
Back
Top