1366 x58 Xeon Enthusiast overclocks club

Intel sadly never made the same "mistake" again...I don't think any of their platforms ever had such longevity.


well...I get your point and I largely agree

but my sig really is my main rig history, and the PIII is still running 24-7 as an audio server here, long since off the internet of course, but still

It is certainly true that they have not made the same mistake since
 
the PIII is still running 24-7 as an audio server here, long since off the internet of course, but still
P3 was better than p4 for a long time, I often wanted to build a dual P3 tualatin setup. I have a magic slocket or whatever that lets me set voltage, with a 1ghz coppermine on an AOpen AX6BC motherboard. I still think about popping in a tualatin and having some fun. My first watercooling kit included a socket A peltier I could modify to mount to it if I really get time. This guy with duals ran crysis https://ancientelectronics.wordpress.com/2016/02/20/upgrading-my-dual-pentium-iii-tualatin-pc/

I remember when I was younger, my first slot 1 PC was from NEC. It had a real crappy 266 MHz celeron and a 66 MHz 440BX board. I used a slotkey adapter to pop in a 733 (I think) celeron, andded another 128MB of SDRAM, a crappy 2x agp card, and felt like a king lol


Xeon W3690 is rock stable, running Linux and overclock by CoreFreq
Nice! I miss the unlocked multiplier of my w3670, glad my motherboard does 200fsb easy though with a x5670 I have plenty of options!
 
Last edited:
...
// cyring, why not overclock in bios?

B/c you don't have to reboot. Just set the multiplier from the desktop ;)

CPU examples:
* Usage: I'm changing clock as needed depending if I'm coding (idle) or building kernels (high)

* Settings: I keep an eye on the Processor features.
Turbo is for instance disabled when Processor is resuming from Suspend to RAM. Then I just re-enable it with shortcut [t]

* I check if the Memory Controller timings are those I left in BIOS.

CoreFreq_Turbo.png
 
Last edited:
I've had an X5650/12 GB DDR3-1600/MSI X58 Pro-E setup sitting around for a long while that cost me all of $100 (just wanted something cheap with SSE4 that would run games that required it), and it's only now that I put it into service since the old Q6600 build I've been trying to move the rest of the family off of had some parts I needed to carry over, mainly the custom water-cooling setup (my XSPC Raystorm blocks are the only things I have around with LGA1366 mounts) and the graphics card.

Well, I've been let down big time, because it seems like a lot of people can hit 200 BCLK with ease on just 1.3V QPI, but this X58 Pro-E is just not having it past 180 BCLK, which cripples the X5650's ability to hit a steady 4.0-4.2 GHz - even at the max 1.35V QPI that's considered the limit for safe 24/7 voltages on Westmere.

There's a YouTube video of someone using the same board, but pumping a whopping 1.47V QPI to hit 200 BCLK, which seems like a great way to kill a 32nm Westmere CPU through electromigration. Maybe if I were overclocking for sport, I'd do it, but I can't risk it on a system that needs to last a while longer for other family members until the 4770K build can be the new hand-me-down in a year or two.

Do these X58 Pro-E boards just suck hard by comparison to most X58 boards or something? Everything I've found about 1st-gen i7/Xeon overclocking hinges on being able to hit at least 200 BCLK stably, like any system should be able to pull it off. It's going past that that should be the hard part.

I'm pretty good on thermal headroom, too; the CPU doesn't even exceed 50C under Cinebench R20 load. IOH runs a bit toasty, but that's apparently normal on these.
 
I remember people complaining that X58 boards were really hit or miss as to whether they could hit 200 and go over or not at least for the first gen parts but like UltraTaco said the CPU is also a limiting factor there as well. I've got a few X58 boards now but to be honest I've not compared them all to see which ones can and can't hit 200+.

Another thing is if you have 6 sticks that also could be why.
 
I just moved back to my X58 and gave the kids my Z77. I still cant believe how smooth this thing is given that I've had it since like 2009.
 
Another thing is if you have 6 sticks that also could be why.
Six 2 GB sticks for 12 GB total, you're right on that. It's partly why the whole CPU/mobo/RAM combo was so cheap, since denser DIMMs hold their value much better.

Higher-density DIMMs would be preferred, but hey, it was part of a $100 bundle and saved me from having to pull RAM from my 4770K build, currently maxed at 32 GB (four 8 GB sticks of DDR3-2400).

I was offered slightly faster Xeons that could have overclocked much better just for having higher multipliers, but at the time I had bought it, even $100 was a lot to spend, especially on such an old platform.

At any rate, I just wanted a gauge on what the silicon lottery was like on Nehalem/Westmere and X58. I know it was all over the place on Haswell, having lost hard with my initial 4670K and returning it for a 4770K under the guise of wanting HT, but really just wanting something that overclocked decently instead of "barely hits 4.2 GHz on custom water cooling with a sizeable bump up in voltage".
 
That's how my old 970 was, it did 4200 with 1.525v. It was a total pig. 1.525 with this one gets me 4600. It used to do all clocks with lower voltage, it seems all of these specter/meltdown mitigations have taken their toll.
 
I'll be joining the club in a few days! Going from an i7 930 to a X5675. Hoping I get a good one to get me to 4.6ghz!
 
After 10.5-years of running on an Intel i7-920 4C 2.6 GHz @ 3.6 GHz Air I am finally upgrading to the Intel Xeon X5690 6C 3.46 GHz and will try for the 4.2 GHz overclock that seems to be pretty standard overclock on Air after spending all of $85 USD on eBay for it.

Getting 2 extra Cores and ~600 MHz more per core seems like a good gain for such a cheap price. I could have gotten an Intel Xeon X5680 6C 3.33 GHz for $40 USD, half price but $40 USD is not a huge difference and perhaps the binning on the X5690 will be better?

I've had problems overclocking that 920 and could only get fully stable 3.6 GHz at only 1.2V (more voltage didn't help). I was able to get 2D stable at 3.8 GHz and 3.7 GHz but never reach the 3.8-4.0 that other people were getting on 920's since I would not reach full 3D video game stability and would get memory corruption and PunkBuster kicked from Battlefield 2 games of that era on anything higher than 3.6. Even 3.65 didn't work for 3D game stability.

PS: Thanks to Dillon from this weekend's party for telling me about the cheap eBay prices on CPUs now for the LGA 1366 on Intel X58 chipset and compatibility between the Intel Core i7-9xx series and Xeon X56xx series.

PPS: This is my first post in 8-years since I've been using the same machine with video card upgrades for STALKER, Oblivion, and Skyrim, SSD and HDD upgrades throughout, and recently a 12 to 24 GB memory upgrade for Fallout 4 high texture pack requirements. This Intel Core i7-920 @ 3.6 GHz X58 chipset and LGA 1366 rig has been rock solid and going strong and this X5690 hopefully will make it better and keep it going. Frankly this one is perfect and quick enough for me that I don't want to build a new rig!!!

bro. I have did the same upgrade month

had my i7 920 max oc at 3.6 and ran ran at below stock speed. and couldn’t get it higher. Upgraded to the x5690 andcurrently running at 3.8 at stick speeds. I haven’t pushed it further because I’m running it at least 15 c cooler than when i has the i7 920. Works much better in a lot of things too
 
Well it's been a while since he posted that, maybe he figured it out.

Vietkangta, in order to 'up dram frequency' to as close as possible to 1600, he would have to increase bclk to 200(he mentioned that) and drop cpu multiplier to 21x in order to maintain same cpu frequency and have that 1600 ram speed.

Theoretically, there should be no problems and voltages don't need any tweaking. In fact, he might even be able to lower them just a tad, because odd multipliers require a little bit less voltage than even. I have witnessed it myself when trying to achieve stable overclock.

Also, regarding his 1600 ram, on regular i7, his uncore would have needed 3200 speed, and that might need little VTT increase, But exons can run as low as 1.5x ram frequency when it comes to uncore vs requiring it be at least 2x like regular i7, so he might even be able to keep that uncore speed as well, no need to change it.

Perhaps he might chime in if he still uses that cpu and configuration.



// PS- he changed his setup, it's on the next page.

I was asking that because I’m currently running the x5690. With bclk of 165 with my multiplier at 23 for cpu to get 3.8ghz

in my ram setting in bios. I was able to set the dram frequency to 1650 (instead of 1320). as one of the dram frequency options on the menu. It was like the 2nd ram option under dram frequency.

Am i doing this wrong? I just hopped on the Xeon crew and somewhat new to ocing
 
Alright, finally an update. I replaced the old i7 920 CPU with a new one X5690 and set the BCLK to 165 from 180 and set the voltage levels in the BIOS back to Auto, which cranked up the CPU to 1.35 V from 1.2 V and the RAM from 1.5 V to 1.56 V which is a bit lot more than the old system running on manually set 1.2 V and 1.5 V but what the hell at this point in time. The new CPU X5690 from stock 3.47 GHZ is at 4.29 GHz (165 MHz x 26-Multi). I haven't run any long-term stability tests but the quick one-run CPU and GPU benchmarks run fine. I set the BCLK to 175 MHz and I get instant blue-screen so I hope that the lower 165 MHz with the higher voltages set by the BIOS Auto settings are fine. I tried running it at 170 MHz but CineBench R20 crashes quickly.

At this point in time I don't have the patience to eek out more performance on the bus clock speed since I'm very happy with 4.29 GHz on Air XigmaTek single 120mm Fan even though I would have been happy with 4.2 GHz.

Not sure if it's stable, will play some 3D games to test since the real test is 3D Gaming and all the other 2D and 3D benchmarks can go to hell for all that I care since I learned that stability is only certain in 3D games such as the old Battlefield 2 (2009) after 15-minutes of no crashes and no PunkBuster kicks.

If this works then I am done! Only things I did was CPU replacement with cleaning off old heat spread and applying new one and changing BCLK to 165 from 180 and voltages setting to Auto for CPU and DRAM. Easiest upgrade ever it seems and cheaper than an Intel Core i7 990X which is going for $250 USD right now on eBay and $120 USD for 980X also.

I played a bit with the memory DIMMs DDR3 at 1600 MHz at 1,654 MHz (2:8) at 1.8 V Auto voltage setting and then at 1,322 MHz (2:10) at 1.56 V Auto voltage and settled on the slower setting since there was no performance increase and the Auto voltage setting bumped the memory up too high in voltage since it used to run stable at 1.5 V for 10-years, and the new 24 GB at ~ 3-years or so. I'd rather have lower voltage on the memory than slightly more speed, especially since at the higher speed the Row Refresh Cycle Time (tRFC) is at 208 clocks at 1,654 MHz (2:10) versus 139 clock on the slower 1,322 (2:8) multi so the performance takes a hit at the higher speed and doesn't yield any faster results. (As we also know from testing memory speeds for decades now.)

Updated: I posted pictures and stability at BCLK 170 MHz but it wasn't stable in CineBench R20, so lowered it to 165 MHz and am testing it now. Same deal as before with this motherboard and the old 920 CPU which would run and work fine at BCLK 190 MHz but wasn't fully 3D long-term stable until I lowered it all the way to 180 MHz. I tried to get that old CPU stable at 187, 185, 183, but No-Go until 180 MHz, no matter the voltage I would push through it.

I did the same ram setting and settled for the higher 1650

memory benchmark shows higher score and also on gta v i get higher fps. I ran the tests multiple times so I’m not imagining it
 
X5690 has higher multiplier range, so it's easier to fine tune exact frequencies you like. You're not doing anything wrong at all. You might actually bring that CPU frequency higher and still be able to work out similar dram speed. Also, you dont **have to** use highest available multi if you can't get wanted speeds. You can drop multipliers and bring everything back up using bclk.

Ok cool. I’m just using this pc and waiting for a worthy upgrade when new cpu are much better lol

been eying every new generation of cpu but the increases in performance just wasn’t enough to warrant a full system upgrade when my system could still chug along well.
 
Well that's great! Certain games benefit greatly from higher ram speed! What I'd do is save those specs and try to work out higher cpu frequency, using the same ram speed.

1650/10multi=165 bclk
If you do 1650/8multi,you will get 206 bclk.

With 206 bclk, you can do 20x multi and see how much voltage you need to gain stability. It equals to about 4.1ghz. That is about 300mhz over your current clocks. If voltages are decent, you could try even more optimal 21x multiplier. That will do you 4.3ghz.

At that point you might be hitting thermal wall, or voltage wall, depending on your chip.

With mine it's 200x21=4.2ghz. I'm using 1.35vcore to feed her, and that gives me 1600 ram. Yours might be better because it's a higher binned model.

Yea i could probably go higher if i did more testing but testing takes hours if not days lol

I’m prettty satisfy with my performance gains right now and i only upgradeD to play around during these covid times. Maybe when i get bored I’ll tinker more. The 4ghz number looks nicer than 3.8. Lol 😋

i might want to add more ram but then I’m not sure what more ram would do. Currently I’m running 16gb (4x4gb) on my ASUS deluxe v2 board. For some reason in the manual of this board shows you can use 4 dimm in triple channel configuration.


C*: Supports four (4) modules inserted into the orange slots (A1, B1 and
C1) and the black slot A2 as one set of Triple-channel memory
configuration.

When i do this my copy speed on aida64 shows drop in performance but other testings are the same. I had 3x4gb originally.


But in other benchmark it didn’t make a difference so I’ just using 16gb


All programs shows my ram in triple channel so there’s that.
 
Well, it's probably some FrankenTriple channel mode. I used one time 2gb sticks, 4gb sticks and 1gb sticks all combined and she would boot, but gave me errors during teste. That was before I found one module to be defective.

Btw, if you want a quick and dirty stability test, cinebench r20 gives a good run for the money. Crashes you right straight if unstable. Once passed, she's fairly good, but of course, further testing and maybe little vcore bump needed to ensure solidness.

Btw, I once did almost 5ghz with like 1.6vcore. It was 2 cores active in bios to compete in superpi😁👍beat all them ryzens!!

Yea I’ve ran cinebench and plus many others just to get the low 3.8ghz lol

the ability to multitask more program is a plus for this new 6 core.
I don’t think I’ll raise my Vcore as high as yours lol. I’m not ready for a new potential oc yet. Lol
 
Have you set your QPI link rate to lowest setting? Those can often put a cap on overclock potential.

yea I’ve set it like that. I haven’t really tested anything beyond 3.8ghz. Basically i inputted some numberS i believe is reasonable and did some test and the machine seems stable. I just don’t have a want to tweak anything right now to get more performance.

i never seen my pc temperate below 35c idle and max below 80c. I kind of like it
 
Last edited:
Death Stranding was on sale a few weeks ago, but looking at the requirements I was surprised to see AVX specifically mentioned. We don't have that instruction set sadly. Are we finally getting to the point where we won't be able to run modern things? I'm getting tired of the random freezing and BSOD's while loading Win10, the codes/descriptions don't line up with the overclocking guide (mostly it just freezes and doesn't log in Event Viewer or make a dump). Sad panda.
 
Death Stranding was on sale a few weeks ago, but looking at the requirements I was surprised to see AVX specifically mentioned. We don't have that instruction set sadly. Are we finally getting to the point where we won't be able to run modern things? I'm getting tired of the random freezing and BSOD's while loading Win10, the codes/descriptions don't line up with the overclocking guide (mostly it just freezes and doesn't log in Event Viewer or make a dump). Sad panda.
your just unstable or something.....i had ZERO freezes and crashes back with my x58 rig for very long durations at a time. With that being said the new x5800 (5x series) look to have some real power and efficiency thats calling my name
 
Death Stranding was on sale a few weeks ago, but looking at the requirements I was surprised to see AVX specifically mentioned. We don't have that instruction set sadly. Are we finally getting to the point where we won't be able to run modern things? I'm getting tired of the random freezing and BSOD's while loading Win10, the codes/descriptions don't line up with the overclocking guide (mostly it just freezes and doesn't log in Event Viewer or make a dump). Sad panda.

Sadly that was the case for me about a month ago, the X5670 running on Gigabyte X58-Extreme at 4.2Ghz + 24GB of RAM was still plenty fast for my needs but the platform was holding it back, specifically in the following ways (mostly related to PCIe):
  1. Lack of PCI-e 3.0 and NVME support, I've researched lots of thread on how to get NVME working but suddenly noticed our lanes are only PCI-e 2.0 and suddenly realized wow is this really that old?
  2. Lack of SATA 6gbps, because of this, I had been running a SAS2.0 RAID card to power my SSDs, this took up yet another precious PCI-e slot
  3. Lack of USB3.0, was able to take advantage of that PCI-e 1x slot with an adapter but yea
It was sad removing and letting the old kit go, I've literally had that motherboard since 2011 and Gigabyte support has been superb.

I am glad to say my new kit is still staying in the lineage - Asus X99 + Xeon E5-2678 v3 + 64GB ECC RAM, while can't truly overclock this I am running the turbo lock mod. Good times :)
 
  • Like
Reactions: bjtag
like this
Sadly that was the case for me about a month ago, the X5670 running on Gigabyte X58-Extreme at 4.2Ghz + 24GB of RAM was still plenty fast for my needs but the platform was holding it back, specifically in the following ways (mostly related to PCIe):
  1. Lack of PCI-e 3.0 and NVME support, I've researched lots of thread on how to get NVME working but suddenly noticed our lanes are only PCI-e 2.0 and suddenly realized wow is this really that old?
  2. Lack of SATA 6gbps, because of this, I had been running a SAS2.0 RAID card to power my SSDs, this took up yet another precious PCI-e slot
  3. Lack of USB3.0, was able to take advantage of that PCI-e 1x slot with an adapter but yea
It was sad removing and letting the old kit go, I've literally had that motherboard since 2011 and Gigabyte support has been superb.

I am glad to say my new kit is still staying in the lineage - Asus X99 + Xeon E5-2678 v3 + 64GB ECC RAM, while can't truly overclock this I am running the turbo lock mod. Good times :)

Well I can't saturate the PCI-e 2.0 bus no matter what I do and I have an LSI PCI-e 8x raid card flashed to IT mode for SATA3 as well as a NVME to PCI-e 8x adapter. Perhaps it's my ageing NVME SSD, but performance is still far better than any SATA3 SSD and I never want for any more power.

Yeah, USB 3.0 would be nice. But TBH, it's not something I in any way lust over - I'm not going to upgrade just for USB 3.0. I'm certainly not counting my use of PCI-e slots. The X99 platform is nice, but it's still demanding a high price as a result.
 
I used mine for 10 or 11 years. We had a good run. I cant bring myself to sell her. I've listed her a few times, and had buyers right away, but I caved with an apology. She is sitting in a box right now. I don't want to power her up because of her incessant need for electricity. The next time she gets power it will be for nostalgia, benchmarking, or because this one died.
 
Well I can't saturate the PCI-e 2.0 bus no matter what I do and I have an LSI PCI-e 8x raid card flashed to IT mode for SATA3 as well as a NVME to PCI-e 8x adapter. Perhaps it's my ageing NVME SSD, but performance is still far better than any SATA3 SSD and I never want for any more power.

Yeah, USB 3.0 would be nice. But TBH, it's not something I in any way lust over - I'm not going to upgrade just for USB 3.0. I'm certainly not counting my use of PCI-e slots. The X99 platform is nice, but it's still demanding a high price as a result.

My SAS2 raid card was perfectly fine in X58 as it was a PCI-e 2.0 based solution, however the speed (esp. sequential) on nvme had a large gap going from PCI-e 2 to 3. Almost like when SSDs first hit the scene with sata2 and maxed it out, then immediately saturated sata3 when it came out.

I also agree that used X99 (or X79 or many other high end platforms) is overpriced. $200ish is what I still see most days. I actually thought about making this move back in 2018 but DDR4 prices were also insane back then.
I was able to find a great price locally for both and really that's one of the main reasons I upgraded
 
L
Sadly that was the case for me about a month ago, the X5670 running on Gigabyte X58-Extreme at 4.2Ghz + 24GB of RAM was still plenty fast for my needs but the platform was holding it back, specifically in the following ways (mostly related to PCIe):
  1. Lack of PCI-e 3.0 and NVME support, I've researched lots of thread on how to get NVME working but suddenly noticed our lanes are only PCI-e 2.0 and suddenly realized wow is this really that old?
  2. Lack of SATA 6gbps, because of this, I had been running a SAS2.0 RAID card to power my SSDs, this took up yet another precious PCI-e slot
  3. Lack of USB3.0, was able to take advantage of that PCI-e 1x slot with an adapter but yea
It was sad removing and letting the old kit go, I've literally had that motherboard since 2011 and Gigabyte support has been superb.

I am glad to say my new kit is still staying in the lineage - Asus X99 + Xeon E5-2678 v3 + 64GB ECC RAM, while can't truly overclock this I am running the turbo lock mod. Good times :)
There was a number of X58 boards that were equipped with SATA 3 and USB 3.0. Granted I think most had the marvel controller which sucked. I ran an X58 Sabertooth for a while and I loved that board. So sad it randomly stopped posting. X58 is still plenty capable for many things today . Just look at how many people are still using an AMD FX platform.
 
L
There was a number of X58 boards that were equipped with SATA 3 and USB 3.0. Granted I think most had the marvel controller which sucked. I ran an X58 Sabertooth for a while and I loved that board. So sad it randomly stopped posting. X58 is still plenty capable for many things today . Just look at how many people are still using an AMD FX platform.

Yep, there was a "refresh gen" of X58 that added USB3 and some had SATA6. Don't get me wrong, for me honestly it was an economics decision, keep putting $ into X58 to get where I want to use the same $ to upgrade platform to something more flexible. I was able to pick up an X99 board for a very reasonable (for me) price so decided to move on.

ChineseStunna mate, what voltage did you have for that 4.2? Taco has 1.35vcore

Another funny(not funny) phenomenon, if I ever push close to 1.5vcore, pc would turn off when I run cinebench. I can hear fans nd pump rpm sag a little and pooff.....OFF everything. My PSU is 750 watts and mobo is asus p6t del v1. Nt sure if vrm's poop or PSU running out of oomph. *ultra shrug*

This does nt occur when using only 1 or 2 cores(other cores disabled in bios). Taco even cn push 1.6voots nd nothing bad.

If I recall correctly about same as you, it was very easy and I didn't spend a ton of time to milk every little bit out of the platform. I went from an i7 950 at 4.2ghz (21x200) to X5670 running the same 21x200FSB settings. I saw some folks got to 4.4ghz but my board had a strange quirk where if I set multi to max of CPU, it would try to Turbo it +1, so I literally had 21x or 23x and not 22x. Pairing with DDR3-1600 the 200 FSB was a happy place.
I think it might be a VRM issue as 750 should be plenty, I have a 750w PSU as well and was even running dual 7970 crossfired just fine.
 
One additional thought I'd like to add is that over the years our priorities/goals change. When I was younger it was literally max performance at cost of everything else: power/heat, noise, stability, PITA (pain in the ass) - built a custom watercooler and loop from Home Depot parts before pre-fabbed kits/parts were available, it was fun. Overclocking is fun and remain my passion, but the amount of heat these chips put out full tilt makes my office a furnace and all the fan noise around it. Nowadays I still have fun tweaking/tuning but not going to spend days doing it as this is my production machine and work needs to get done, has to be rock solid stable and reasonably low heat/noise output.

I think current platform X99 + E5-2678 v3 is actually slower than X58 + X5670 @ 4.2ghz in single threads, as the 2678 will only do 3.3ghz max. I debated getting a Xeon 1660 v3 to overclock and get the same performance as X5670 but again, economics - that chip is 6 core @ about $170 on eBay vs. 2678 is 12 cores @ $80. Multithreaded performance with turbo unlock is about DOUBLE (cinebench R15, even more with AVX loads) vs. X5670 all while using a lot less power.
 
Last edited:
My SAS2 raid card was perfectly fine in X58 as it was a PCI-e 2.0 based solution, however the speed (esp. sequential) on nvme had a large gap going from PCI-e 2 to 3. Almost like when SSDs first hit the scene with sata2 and maxed it out, then immediately saturated sata3 when it came out.

My LSI raid card flashed to IT mode is only used for spinners, because I run Linux I have one 6TB HDD set up as /home and one 6TB HDD set up as a storage drive for general crap. Using mechanical spinners there's no way I can saturate that bus.

My NVME to PCI-e 8x SSD is used for the root file system, even that doesn't saturate the bus and benchmarks far faster than any SATA3 SSD. My system's quiet (compared to my overclocked machines of the past it's dead silent!) and with a 75w TDP my 2 x X5675's actually aren't that bad regarding power and heat output, in fact I never turn my PC off as I can apply 90% of updates without needing a reboot.

Your comments regarding X99 in single threaded application is interesting, the slower clock speeds in favor of core count it's one of the reasons I've held off upgrading.
 
My LSI raid card flashed to IT mode is only used for spinners, because I run Linux I have one 6TB HDD set up as /home and one 6TB HDD set up as a storage drive for general crap. Using mechanical spinners there's no way I can saturate that bus.

My NVME to PCI-e 8x SSD is used for the root file system, even that doesn't saturate the bus and benchmarks far faster than any SATA3 SSD. My system's quiet (compared to my overclocked machines of the past it's dead silent!) and with a 75w TDP my 2 x X5675's actually aren't that bad regarding power and heat output, in fact I never turn my PC off as I can apply 90% of updates without needing a reboot.

Your comments regarding X99 in single threaded application is interesting, the slower clock speeds in favor of core count it's one of the reasons I've held off upgrading.
Interesting you have a PCI-e 8x nvme card, I'm guessing it's an enterprise unit? Curious which model you went with?
Again I'm not super thrilled with the X99 solution but it's the best option given my requirements (budget/multithread/power/noise)
 
Interesting you have a PCI-e 8x nvme card, I'm guessing it's an enterprise unit? Curious which model you went with?
Again I'm not super thrilled with the X99 solution but it's the best option given my requirements (budget/multithread/power/noise)
Actually....Sorry, it's an m.2 to PCI-e adapter and it's x4 - But my LSI raid card is definitely x8. Performance still smashes any SATA3 SSD out there and the PCI-e 2.0 bus doesn't really seem to be a bottleneck.

 
Actually....Sorry, it's an m.2 to PCI-e adapter and it's x4 - But my LSI raid card is definitely x8. Performance still smashes any SATA3 SSD out there and the PCI-e 2.0 bus doesn't really seem to be a bottleneck.


Ah that makes sense then, I was looking into similar solutions as well. Glad you got it working especially as the boot device (read all about Clover/efi emulation).
My LSI raid card was fine too driving a 4x SSD R0 array and 2x SSD r0 array, the upgrade to X99 with native SATA3 ports allowed me to run those off the mobo directly and seems to actually gained a bit of performance according to benchmarks (nothing noticeable). The LSI RAID card is now driving 8X SAS 10k drives for a R5 storage array
 
Ah that makes sense then, I was looking into similar solutions as well. Glad you got it working especially as the boot device (read all about Clover/efi emulation).
My LSI raid card was fine too driving a 4x SSD R0 array and 2x SSD r0 array, the upgrade to X99 with native SATA3 ports allowed me to run those off the mobo directly and seems to actually gained a bit of performance according to benchmarks (nothing noticeable). The LSI RAID card is now driving 8X SAS 10k drives for a R5 storage array

Sorry about that, I fitted the card years ago and confused the interface with my LSI raid card. Yes, the device is bootable - It's actually a very effective solution and wasn't that expensive at the time.
 
I've still got one 1366 system in active use. My work uses my hand-me-down as a file and domain server.
 
your just unstable or something.....i had ZERO freezes and crashes back with my x58 rig for very long durations at a time. With that being said the new x5800 (5x series) look to have some real power and efficiency thats calling my name

Yup, and this has been stable for years too. My 920 overclock degraded from 4.2 to 3.8 before I went x5670, and this has been going at 4.0 for years (I could only get 4.2 anyway). At some point hardware just goes. PSU is a 10-11yr old Corsair HX1000, mobo is 10, etc.
 
Well after probably 2-3 years of infrequent freezing at Win10 login screen, and only occasionally giving me a blue screen during those freezes, it's finally made a couple minidump files. Code 0x9C, it could have told me earlier but nooooo. "0x9C = QPI/VTT most likely, but increasing vcore has helped in some instances"

My QPI was already at the advised max of 1.35V, so I upped my Vcore from 1.35 to 1.375V. Fingers crossed!
 
Last edited:
I'm thinking of getting either a W3680 or X5675 to upgrade my trusty old i7 920 which sits in a Gigabyte X58 Extreme motherboard. The W3680 has an unlocked multiplier but also has a higher TDP when compared to the X5675. I can get both for the same price and would like to overclock the CPU, wondering if anyone can shed advice on which one to get?
 
Back
Top