Who ran Haswell-E for a long time or is still running it?

Joined
Apr 9, 2021
Messages
943
I'm looking for some long-term overclocking data for the 5820K, 5930K, 5960X, E5-1650V3, E5-1660V3, and E5-1680V3, now that the chips are actually old enough for long-term data to exist.

What did you overclock your chip(s) to (Core/Cache/DRAM)? What voltages (vcore, vring, vccin, vccsa, vccio)? Cooling? Motherboard? How long did you run it? How was it used? How did it hold up?
 
I run dual E5-2667v3

While old, this still stacks up pretty well. For example, vs an i9-10900 (non-K)
z840vs3640.png


So you see the much higher single core that you'd expect being many generations ahead, but you still have the greater multi-core due to having 16 real cores vs the 10 in the i9.

I think the big differentiator has to do with PCIe channels and amount of memory that can be handled. You're not going to be able to put over 1TB of memory on a desktop CPU (maybe not ever).

My z8640 has 160G of memory for example. While you might be able to put 128G on some TR setups (for example), usually you start getting into performance trouble beyond that. Not that it still wouldn't stand ok against the old Z840, but enthusiasts will be sad as they'll clock way down for stability. There's a reason you don't see 256G TR setups often.

In short, an old Haswell dual (I said dual!) will do quite nicely today. While it might not be as nice as some of the new and shiny, it's still fairly competitive and might come at a price that's hundreds of dollars less and might be better in other ways, memory size, cooling and reliability.

(sorry about the GB version difference, I don't think it make a huge difference on the values presented)

Now for PassMark. Note, I did these on the fly. The Z840 is my primary workstation. While not terribly heavily loaded, it was running a couple of browsers, a few application, several VMs, etc.. The 3640 was pretty idle.

PassMark PerformanceTest Linux Intel(R) Xeon(R) CPU E5-2667 v3 @ 3.20GHz (x86_64) 16 cores @ 3600 MHz | 157.3 GiB RAM Number of Processes: 32 | Test Iterations: 1 | Test Duration: Medium -------------------------------------------------------------------------- CPU Mark: 24343 Integer Math 97802 Million Operations/s Floating Point Math 48180 Million Operations/s Prime Numbers 118 Million Primes/s Sorting 57104 Thousand Strings/s Encryption 9269 MB/s Compression 415 MB/s CPU Single Threaded 1971 Million Operations/s Physics 1480 Frames/s Extended Instructions (SSE) 21954 Million Matrices/s Memory Mark: 2448 Database Operations 10574 Thousand Operations/s Memory Read Cached 23906 MB/s Memory Read Uncached 9392 MB/s Memory Write 8664 MB/s Available RAM 125205 Megabytes Memory Latency 54 Nanoseconds Memory Threaded 61571 MB/s --------------------------------------------------------------------------

PassMark PerformanceTest Linux Intel(R) Core(TM) i9-10900 CPU @ 2.80GHz (x86_64) 10 cores @ 5200 MHz | 31.1 GiB RAM Number of Processes: 20 | Test Iterations: 1 | Test Duration: Medium -------------------------------------------------------------------------- CPU Mark: 20784 Integer Math 83525 Million Operations/s Floating Point Math 51870 Million Operations/s Prime Numbers 66 Million Primes/s Sorting 34775 Thousand Strings/s Encryption 8823 MB/s Compression 277 MB/s CPU Single Threaded 3293 Million Operations/s Physics 1091 Frames/s Extended Instructions (SSE) 13847 Million Matrices/s Memory Mark: 3815 Database Operations 8396 Thousand Operations/s Memory Read Cached 35071 MB/s Memory Read Uncached 17930 MB/s Memory Write 17052 MB/s Available RAM 29027 Megabytes Memory Latency 29 Nanoseconds Memory Threaded 37041 MB/s --------------------------------------------------------------------------
 
All of these were run in an ASUS X99-A, just not the same one. Cooling for all was a Phanteks TC14PE air cooler and just recently switched to an EVGA CLC 280.

5820K #1
Core/Cache/RAM: 4.5/4.4/2400
Voltage: 1.28V/1.25/1.9/stock/stock
Lasted: 1 month
Cause of Death: ASUS Overvoltage glitch. Motherboard and CPU dead. Code 00.

5820K #2
Core/Cache/RAM: 4.5/4.5/2400
Voltage: 1.25V/1.25/1.9/stock/stock
Lasted: 3 months
Cause of Death: ASUS Overvoltage glitch. Motherboard and CPU dead. Code 00

5820K #3

Core/Cache/RAM: 4.6/4.4/2666
Voltage: 1.22V/1.20/1.9/0.9/stock
Lasted: 1 year
Cause of Death: Not known for sure. Motherboard and CPU were both dead. Code B2

5820K #4
Core/Cache/RAM: 4.2/3.8/2400
Voltage: 1.22V/1.12V/1.85/stock/stock
Lasted: 4 years, still works, but sits in a drawer. Murphy's Law of Overclocking: The duds never seem to die.

5960X (5 years old when I bought it, unknown history)
Core/Cache/RAM: 4.2/4.0/2666
Voltage: 1.17V/1.08V/1.88/stock/stock
Lasted: Still kicking.
 
Last edited:
I'm looking for some long-term overclocking data for the 5820K, 5930K, 5960X, E5-1650V3, E5-1660V3, and E5-1680V3, now that the chips are actually old enough for long-term data to exist.

What did you overclock your chip(s) to (Core/Cache/DRAM)? What voltages (vcore, vring, vccin, vccsa, vccio)? Cooling? Motherboard? How long did you run it? How was it used? How did it hold up?
I ran my 5960x J-Batch Haswell-e at 4.625Ghz at 1.31Vcore solid, never an issue and is still running that way today in my entertainment Center PC (upgraded my gaming PC to AMD 5950x).

I did not have an OC socket on my AsRock X99 OC Formula board, so Cache was limited to 3.6Ghz for me (I think this is where most people killed Haswell-e CPUs honestly for minimal gains).

Dram was Samsung B-Die, 32GB (4x8GB) and was able to run it at 3000Mhz 13-13-13-26 @1T with 1.34V. Anything near 3200Mhz+ was a no post no matter the timings, so a decent IMC, but not the best ever for Haswell-e. I know the ram can do it because I moved it to my current build and get 3740Mhz at CAS 16... lol. I threw my original Hynix 16GB kit in my X99 for it's use as a TV PC now and it runs at 2800Mhz CAS 14 (original DDR4 sticks from 2014).

I think i had my Vccin at 1.78V, vccsa at 1.15V and Ring at 1.05V. I'd have to check on all of that though to be sure, it's been a little while.

I cooled the CPU with a Corsair H100i XT Pro (upgraded from original H100i at some point) and it did a great job, temps always under 75C full load.

But it was a damn good PC; went through 4 video card generations with that build and even had my 3090 in it for a few months and it worked great for gaming, although I get no stutter now and would occasionally on the 5960x in CPU heavy games that are out now, so it was time for a new build.
 
I run dual E5-2667v3

While old, this still stacks up pretty well. For example, vs an i9-10900 (non-K)
View attachment 374891

So you see the much higher single core that you'd expect being many generations ahead, but you still have the greater multi-core due to having 16 real cores vs the 10 in the i9.

I think the big differentiator has to do with PCIe channels and amount of memory that can be handled. You're not going to be able to put over 1TB of memory on a desktop CPU (maybe not ever).

My z8640 has 160G of memory for example. While you might be able to put 128G on some TR setups (for example), usually you start getting into performance trouble beyond that. Not that it still wouldn't stand ok against the old Z840, but enthusiasts will be sad as they'll clock way down for stability. There's a reason you don't see 256G TR setups often.

In short, an old Haswell dual (I said dual!) will do quite nicely today. While it might not be as nice as some of the new and shiny, it's still fairly competitive and might come at a price that's hundreds of dollars less and might be better in other ways, memory size, cooling and reliability.

(sorry about the GB version difference, I don't think it make a huge difference on the values presented)

Now for PassMark. Note, I did these on the fly. The Z840 is my primary workstation. While not terribly heavily loaded, it was running a couple of browsers, a few application, several VMs, etc.. The 3640 was pretty idle.

PassMark PerformanceTest Linux Intel(R) Xeon(R) CPU E5-2667 v3 @ 3.20GHz (x86_64) 16 cores @ 3600 MHz | 157.3 GiB RAM Number of Processes: 32 | Test Iterations: 1 | Test Duration: Medium -------------------------------------------------------------------------- CPU Mark: 24343 Integer Math 97802 Million Operations/s Floating Point Math 48180 Million Operations/s Prime Numbers 118 Million Primes/s Sorting 57104 Thousand Strings/s Encryption 9269 MB/s Compression 415 MB/s CPU Single Threaded 1971 Million Operations/s Physics 1480 Frames/s Extended Instructions (SSE) 21954 Million Matrices/s Memory Mark: 2448 Database Operations 10574 Thousand Operations/s Memory Read Cached 23906 MB/s Memory Read Uncached 9392 MB/s Memory Write 8664 MB/s Available RAM 125205 Megabytes Memory Latency 54 Nanoseconds Memory Threaded 61571 MB/s --------------------------------------------------------------------------

PassMark PerformanceTest Linux Intel(R) Core(TM) i9-10900 CPU @ 2.80GHz (x86_64) 10 cores @ 5200 MHz | 31.1 GiB RAM Number of Processes: 20 | Test Iterations: 1 | Test Duration: Medium -------------------------------------------------------------------------- CPU Mark: 20784 Integer Math 83525 Million Operations/s Floating Point Math 51870 Million Operations/s Prime Numbers 66 Million Primes/s Sorting 34775 Thousand Strings/s Encryption 8823 MB/s Compression 277 MB/s CPU Single Threaded 3293 Million Operations/s Physics 1091 Frames/s Extended Instructions (SSE) 13847 Million Matrices/s Memory Mark: 3815 Database Operations 8396 Thousand Operations/s Memory Read Cached 35071 MB/s Memory Read Uncached 17930 MB/s Memory Write 17052 MB/s Available RAM 29027 Megabytes Memory Latency 29 Nanoseconds Memory Threaded 37041 MB/s --------------------------------------------------------------------------
Note: My post wasn't really about OC, more about viability of the old Haswell vs something new.
 
I ran my 5960x J-Batch Haswell-e at 4.625Ghz at 1.31Vcore solid, never an issue and is still running that way today in my entertainment Center PC (upgraded my gaming PC to AMD 5950x).

I did not have an OC socket on my AsRock X99 OC Formula board, so Cache was limited to 3.6Ghz for me (I think this is where most people killed Haswell-e CPUs honestly for minimal gains).

Dram was Samsung B-Die, 32GB (4x8GB) and was able to run it at 3000Mhz 13-13-13-26 @1T with 1.34V. Anything near 3200Mhz+ was a no post no matter the timings, so a decent IMC, but not the best ever for Haswell-e. I know the ram can do it because I moved it to my current build and get 3740Mhz at CAS 16... lol. I threw my original Hynix 16GB kit in my X99 for it's use as a TV PC now and it runs at 2800Mhz CAS 14 (original DDR4 sticks from 2014).

I think i had my Vccin at 1.78V, vccsa at 1.15V and Ring at 1.05V. I'd have to check on all of that though to be sure, it's been a little while.

I cooled the CPU with a Corsair H100i XT Pro (upgraded from original H100i at some point) and it did a great job, temps always under 75C full load.

But it was a damn good PC; went through 4 video card generations with that build and even had my 3090 in it for a few months and it worked great for gaming, although I get no stutter now and would occasionally on the 5960x in CPU heavy games that are out now, so it was time for a new build.

I think you're right about the OC socket. All my dead HW-E chips were running big cache overclocks. My guess is that the BIOS is programmed to juice those OC socket pins with more voltage when the cache OC is higher, and too much of that makes the FIVR go bananas. I've noticed that the BIOS versions that allow the highest cache clocks with the least amount of ring voltage are the ones that caused me the most problems.
 
Note: My post wasn't really about OC, more about viability of the old Haswell vs something new.

It's all good, I agree it's still pretty viable. My favorite thing about the platform was that it was the best at literally everything when it came out. Professional work or gaming, it didn't matter. The current X299 is great for productivity, but it gets its ass kicked in gaming by the mainstream Intel and AMD platforms. Actually, the 5950X probably kicks its ass at just about everything, just about all the time - except for the fact that the AM4 platform is still hit or miss with large memory sizes, and limited in PCI-E lanes.
 
I think you're right about the OC socket. All my dead HW-E chips were running big cache overclocks. My guess is that the BIOS is programmed to juice those OC socket pins with more voltage when the cache OC is higher, and too much of that makes the FIVR go bananas. I've noticed that the BIOS versions that allow the highest cache clocks with the least amount of ring voltage are the ones that caused me the most problems.
Yeah, it was a fairly well known problem with the OC Socket boards for x99. There was little data from Intel on safe Cache voltage ranges, but many people warned never to go above 1.15V (Ring) and anything at 1.25V or higher was flirting with death. The worst part was, know one really knows what the OC socket did as the voltage it applied was a big black box on how it all worked.

Another CPU killer was SA voltage and people trying to get higher DDR4 Ram overclocks. Many people put 1.25V or higher on the IMC and chips died a year or less later. Safe range for SA voltages on Haswell-e was 1.15V or less, but a lot of people pushing 3200Mhz+ on the DDR4 needed well over that for stability.

I think it was a combination of SA and Ring voltages killing those chips for people. I kept both of mine below "established" safe limits and never had any issues on my 5930k or 5960x.... both work to this day with overclocked settings.
 
Yeah, it was a fairly well known problem with the OC Socket boards for x99. There was little data from Intel on safe Cache voltage ranges, but many people warned never to go above 1.15V (Ring) and anything at 1.25V or higher was flirting with death. The worst part was, know one really knows what the OC socket did as the voltage it applied was a big black box on how it all worked.

Another CPU killer was SA voltage and people trying to get higher DDR4 Ram overclocks. Many people put 1.25V or higher on the IMC and chips died a year or less later. Safe range for SA voltages on Haswell-e was 1.15V or less, but a lot of people pushing 3200Mhz+ on the DDR4 needed well over that for stability.

I think it was a combination of SA and Ring voltages killing those chips for people. I kept both of mine below "established" safe limits and never had any issues on my 5930k or 5960x.... both work to this day with overclocked settings.

I never went nuts with the RAM because my sticks won't go faster than 2666/C14, but I did push the cache, and things went haywire every time. Maybe I could be a little more aggressive with the cores on my current chip. It's a J-batch, and it clocks like a J-batch.

The funny thing is that I also have a 4690K, and I've been running its cache at 4GHz/1.2V for 6 years without the slightest amount of degradation. Actually, the default cache speed/voltage on the chip is [email protected], even though the stock all-core turbo is actually lower at only 3.7/1.095V. The ring bus seems happier on 4 cores than 6/8/10.
 
It's not my primary system anymore, but my 5820k system is still running like a champ for my 1080p HDTV in the family room. It's got a 1660 Super paired with. Way more than enough for 1080p gaming still. I've been running this thing overclocked @ 4ghz for years. I used to run it higher, but didn't see the need/point once it was only doing 1080p/60hz gaming.
 
I'm still running my 5820K @ 4.5Ghz. 1.25v. Ram is Corsair 3000Mhz, C15, 4x 4GB modules in quad-channel. Motherboard is "EVGA X99 FTW". I didn't do any advanced cache overclocking but I do use LLC.

The system is still running great. It was replaced as my main system by my 3900X about a year and a half ago. It functions as my backup system now, on a KVM, attached to all 3 of my monitors. I run it now with 3x GTX680 in Triple-SLI to take advantage of the large number of PCIe lanes, and turn these 3 old GPUs into something that can still pump out good FPS when I need to use the system (all the games I care about still support DX11 SLI).

The only "issue" I have is that the CM Hyper212 Evo (even with 2 high-CFM fans in push-pull) is not really adequate for the 4.5Ghz overclock. It can barely complete a Cinebench run without giving me a BSOD, but for actual gaming it's still solid at those clocks. I ran it with an AIO when it was my main system but I didn't want to deal with water cooling in a backup computer.

5820k_4500.png
 
^ Nice. (y)

You guys inspired me to crank mine up a little...
 

Attachments

  • Super_PI_32M_424.878.jpg
    Super_PI_32M_424.878.jpg
    400 KB · Views: 0
You all motivated me to actually post a picture... lol. My AMD 5950x system blows this out of the water in IPC, L1, 2 & 3 performance and raw clocks; but damn I had my x99 5960x signing and still does as my secondary PC on my entertainment center as I stated before. :)

5960x_cache35mem_3000_TestsON_32GB_20_3rd_level_tight.png
 
I have E5-4669v3 (QS) 18-core Haswell on Asrock X99M motherboard. Been running solid for 4 years running various BOINC CPU projects.

Also, all 18-cores are turbo to the max clock i.e. 2.9GHz running non-avx workload. Posted this before in this thread.

Haswell prices have dropped a lot since I last checked. I need more cores!

1626311765469.png
 
You all motivated me to actually post a picture... lol. My AMD 5950x system blows this out of the water in IPC, L1, 2 & 3 performance and raw clocks; but damn I had my x99 5960x signing and still does as my secondary PC on my entertainment center as I stated before. :)

Now you've forced me to do something drastic 🤣

5960X at 4.75.jpg
 
Last edited:
I have E5-4669v3 (QS) 18-core Haswell on Asrock X99M motherboard. Been running solid for 4 years running various BOINC CPU projects.

Also, all 18-cores are turbo to the max clock i.e. 2.9GHz running non-avx workload. Posted this before in this thread.

Haswell prices have dropped a lot since I last checked. I need more cores!

View attachment 375099

Impressive (y) I wish my board supported the E5-4xxx series.
 
Now you've forced me to do something drastic 🤣

View attachment 375114
Assuming that's 100% stable and you can manage to keep it cool with 1.36V, thats an even more impressive overclock than what I had! Those J Batch 5960x were overclocking kings! Probably one of my favorite platforms I have ever messed with for tweaking and overclocking.

Careful with that cache though and the OC socket, don't wanna fry a good core overclocker... lol.
 
I still have my 5820k system up and running. Asus Rampage V Edition 10 motherboard. 32GB RAM.

Ran it with a 39% overclock for a LONG time, but it's running stock now. I currently do everything with it except game. Running Arch Linux for desktop, storage, plex, and a host of other things.

Always ran it on air; Noctua NH-D15.

I originally used it for gaming. Had a 1080GTX x2 SLI setup for about 4 years; played World of Warcraft almost exclusively through Legion and most of BFA.

I've tossed my 2070 super in it a couple of times and haven't noticed much difference between it and my 3600 AMD system when playing FFXIV or SWTOR. The AMD 3600 will run about 30% - 35% utilization and the 5820k will run about 45% utilization in those games at 4k/max settings. Obviously at 4k the games are GPU bound.
 
Assuming that's 100% stable and you can manage to keep it cool with 1.36V, thats an even more impressive overclock than what I had! Those J Batch 5960x were overclocking kings! Probably one of my favorite platforms I have ever messed with for tweaking and overclocking.

Careful with that cache though and the OC socket, don't wanna fry a good core overclocker... lol.

I would not call that 100% stable... just enough to run benches. Cache could have been at 4.5, but I won't do that even for a quick run. I normally don't set it higher than [email protected], but for this I juiced it with 1.16V. :)
 
Last edited:
I've been running a 5820k on water since August 2015. Board has been an ASROCK X99 Extreme 4/3.1.

4.5ghz core 4.1ghz cache

VCCIN = 1.900v
Vcore = 1.35v
Vcache = 1.30v

VCCSA/VCCIO I ran on auto with a terrible kit of 4x4gb ddr2400 for the first ~3 years.
Past 2.5ish years I've had it at I think 1.05 VCCSA and similar VCCIO with 4x8gb CJR @ DDR3200 14-17-17-33.

I recently started getting some BSODs and backed down to 4.4ghz, but that's likely because my full load temps are hitting over 80c now for some reason. I suspect my water block i plugged up with junk, since my flow rate seems terrible now as well.

My 5900x arrived today and is just waiting for my Heatkiller IV Pro AM4 bracket kit to be delivered. X99 did me well, but 6 cores isn't enough anymore and the CPU in general is starting to be a bit of a bottleneck in some games. The final nail in the coffin for me was when I upgraded to Windows 11 and took an ~8% performance hit in CPU benchmark scores (I suspect due to spectre/meltdown mitigations).
 
I run a early haswell es 10 core 10 thread xeon in my desktop. Its at 3.0ghz but can do 3.3 if your feeling frisky :p
 
I got my 5820 in November 2014. Had it up to 4.6 but my motherboard has some power issues so I had to pull it back to 3.8 with nothing special.
 
My office PC I have a 5820k I built on release with some 2667 c14 hyperx 4gb sticks on Asrock X99M Killer. Originally ran a 4ghz simple oc but bumped it up to 4.4 2 years ago or so, got win 8.1 on it but no issues with it at all.
 
Last edited:
My 5960X is still rocking at 4.4Ghz at 1.25v. It runs as a back up rig on the home LAN and now cooled with a Corsair AIO. Back in the day it was on a custom loop at 4.6ghz. Its still a strong rig that crushed video edit, and paired with a 1080 Ti games very well.
 
My 5960X is still rocking at 4.4Ghz at 1.25v. It runs as a back up rig on the home LAN and now cooled with a Corsair AIO. Back in the day it was on a custom loop at 4.6ghz. Its still a strong rig that crushed video edit, and paired with a 1080 Ti games very well.

It's nice to know that they can take the "abuse" long term. Even though I plan on upgrading soon, this CPU will be useful for a very long time, so I don't want to kill it. Had some time to play with it today and see what the max stable (but not Prime 95 stable) OC is. It took a little more voltage than I was hoping it would, but it looks like it can reach back and hurl 4.6 if I really need it to. Otherwise, it lives its life on easy street at [email protected] Prime 95 full-virus-mode stable. (y)

CINEBENCH_R20_CPU_4288.jpg
 
My E5-4669 v3 scores 5500 in R20 after the second run. All cores running at 2827 Mhz with max temp of 57°C during the bench test. Note I'm using a hack BIOS to turbo all the cores within the power limit of the chip (135W). According to this site, it puts E5-4669 v3 R20 score higher than the 24-core Xeon Platinum 8160 and below the 2021 8-core i7-11700K. Not bad for a 6-year old technology.



2021-07-16 17_01_52-BenchMate 0.10.7.2 _ Verified Result.png
 
I'm looking for some long-term overclocking data for the 5820K, 5930K, 5960X, E5-1650V3, E5-1660V3, and E5-1680V3, now that the chips are actually old enough for long-term data to exist.

What did you overclock your chip(s) to (Core/Cache/DRAM)? What voltages (vcore, vring, vccin, vccsa, vccio)? Cooling? Motherboard? How long did you run it? How was it used? How did it hold up?
I had two Haswell-E CPU's I used regularly. Both were Core i7 5960X's I had one on my test bench that would do 4.5GHz @ 1.35v (ES) and another (retail) that would do 4.4GHz @ 1.28v. The latter was in my personal machine while the former was a review CPU that I beat the shit out of on the test bench through the entirety of HardOCP's reviews on the X99 platform. The second chip was run 24/7 @ 4.4GHz its entire service life which was over 4.5 years. After HardOCP went into mothballs I used the 4.5GHz chip in my main rig and then moved it to my girlfriend's machine where it ran 24/7 for about six months before I upgraded both machines.

RAM was always DDR4 3000MHz or better on the review sample and in my personal machine the configuration changed a few times. I do believe it was mostly run at DDR4 2666MHz initially, but eventually I upgraded to DDR4 3200MHz RAM. At the end I think I ran 64GB of DDR4 3000MHz memory on it.
 
My HP Z840 2 x E5-2667v3 (no OC or anything special). However, this is a Windows VM, not bare metal.

cb-win10-vm.png
 
I still have a 5960x in my father's computer. I asked him if he wants me to replace anything, and he tells me not to touch it...
 
My 5960x is still going strong in my secondary system. I am just annoyed it won't be allowed to upgrade to Windows 11.
 
My 5960x is still going strong in my secondary system. I am just annoyed it won't be allowed to upgrade to Windows 11.
I wouldn't take bets on that. I'm betting that really really old stuff will be supported.
 
I am just annoyed it won't be allowed to upgrade to Windows 11.

I'm running the Insider Build on a 15 year old Q6600 right now and it works great. Unless there is a fundamental shift in how Windows 11 works between now and release, it will be trivial for your average enthusiast to install it on anything they want to. I fully intend to run it on my 5820k.
 
Last edited:
I had two Haswell-E CPU's I used regularly. Both were Core i7 5960X's I had one on my test bench that would do 4.5GHz @ 1.35v (ES) and another (retail) that would do 4.4GHz @ 1.28v. The latter was in my personal machine while the former was a review CPU that I beat the shit out of on the test bench through the entirety of HardOCP's reviews on the X99 platform. The second chip was run 24/7 @ 4.4GHz its entire service life which was over 4.5 years. After HardOCP went into mothballs I used the 4.5GHz chip in my main rig and then moved it to my girlfriend's machine where it ran 24/7 for about six months before I upgraded both machines.

RAM was always DDR4 3000MHz or better on the review sample and in my personal machine the configuration changed a few times. I do believe it was mostly run at DDR4 2666MHz initially, but eventually I upgraded to DDR4 3200MHz RAM. At the end I think I ran 64GB of DDR4 3000MHz memory on it.

That ES chip sounds like a real winner. 64MB at 3000 is nothing to sneeze at with these chips. You know, one thing I learned from playing with a lot of a different Haswell-E samples was to stop looking for chips that can run higher clocks on lower voltages. I have 2 5960X's right now, the one in my main machine and another one sitting in a drawer. The chip in the drawer does 4.6 at 1.25, but tbh, it's a real piece of shit. If I run Cinebench on it at 4.6, it'll hit 90 on 2 cores, 80+ on 2 other cores, and 70+ on the other 4. OTOH, the chip I'm using needs 1.33V to be stable at 4.6, but it only runs at 66-72C on all 8 cores and it scales upwards from there very well.
 
That ES chip sounds like a real winner. 64MB at 3000 is nothing to sneeze at with these chips. You know, one thing I learned from playing with a lot of a different Haswell-E samples was to stop looking for chips that can run higher clocks on lower voltages. I have 2 5960X's right now, the one in my main machine and another one sitting in a drawer. The chip in the drawer does 4.6 at 1.25, but tbh, it's a real piece of shit. If I run Cinebench on it at 4.6, it'll hit 90 on 2 cores, 80+ on 2 other cores, and 70+ on the other 4. OTOH, the chip I'm using needs 1.33V to be stable at 4.6, but it only runs at 66-72C on all 8 cores and it scales upwards from there very well.
I can't recall the temps exactly, but I do remember that chip seeming to have more thermal headroom. Unfortunately, when you moved up to 4.6GHz, it required way too much voltage to reach that clock speed and it ran too hot.
 
Does any have a link for the latest x99 chipset drivers?
Your best bet is to use the Intel driver utility in Chrome or Edge. It will detect the drivers. Also you can use the built-in Windows drivers that will download. That is why I use on my secondary machine except graphics drivers.
 
^
I've just been using the latest ones on ASUS's driver page for my board. They're probably not the newest, but everything works fine so w/e.
 
I run my 5930K since the launch at 4.2GHz with 1.280V on VCORE, never succeded in getting a stable 4.3GHz but at 4.2GHz the processor runs great since years.
very disappointed by the fact that this CPU will not run Windows 11
 
I run my 5930K since the launch at 4.2GHz with 1.280V on VCORE, never succeded in getting a stable 4.3GHz but at 4.2GHz the processor runs great since years.
very disappointed by the fact that this CPU will not run Windows 11

Yeah, I've never seen a platform age as well as this one has. It turns out that my 5960X really likes the DDR4 thats available today. I was going to use it for something else, but I ended up installing an 8x8GB 3733C17 rated kit, and it worked straight-away at 3200MT/S, 14-14-14-28 timings at 1.35V and stock 0.81V system agent and stock 1.05V vccio. Tightened up all the secondary timings too and passed 20,000% in Karhu. Nice surprise, since my old Crucial Ballistix would not do above 2666 at any voltages.

gskill.jpg
 
Back
Top