Overclocking Xeon X5482
Overclocking Xeon X5482
Hello everyone.
I must admit I haven’t really dived into overclocking much, even though I once had some plans. My goal is simply to clarify my thoughts.
Recently, I bought a used Intel quad-core Xeon X5482 processor (3.2 GHz, 12MB cache, 1600 MHz FSB, SLANZ-150W max TDP) from the Harpertown line. We all know these Xeons are compatible with LGA771 sockets, but lately there’s been a trend to adapt them for LGA775 boards. This requires a specific chipset—Intel P45-P43-P35-P31-G45-G43-G41-G35-G33-G31 nForce models or GeForce 9400/9300. Boards lacking these chipsets won’t support the Xeon in their BIOS. More information is available on links like this one: http://www.delidded.com/lga-771-to-775-adapter/.
I installed it on a newer motherboard with S775, the Gigabyte GA-EP45T-UD3R. It features DDR3 slots (up to 16GB total, up to 2200+ MHz), and I’m using Corsair Vengeance DDR3 CMZ16GX3M4X1600C9 with XMP profiles. This board lets me fine-tune CPU settings via Gigabyte’s BIOS feature called “M.I.T. – Motherboard Intelligent Tweaker.” This tool lets me adjust parameters like:
- CPU Clock Ratio (usable when unlocked; my clock is locked)
- Fine CPU Clock Ratio (micro-adjustment from 8x to 8.5x)
- CPU Host Clock Control (enables/disables control over frequency)
- CPU Host Frequency (set manually, requires 400 MHz for normal operation)
- CPU Clock Drive (adjusts amplitude and North Bridge from 700mV to 1000mV)
- CPU Clock Skew (sets clock before North Bridge, 0–750ps default)
- MCH Clock Skew (sets clock before CPU, 0–750ps default)
- Load-Line Calibration (affects Vdroop, keeps voltage stable under load)
- CPU Vcore (auto or fixed at 1.2V)
- CPU Termination (default auto; unclear purpose)
- CPU PLL (controls performance based on load, 5 modes)
- C.I.A. 2 (Intel CPU Accelerator 2, adjusts frequency up to 19% more)
To sum up:
1) Which parameters can I choose? Which should I keep or change?
2) For moderate overclocking (3.2 GHz → 3.6–3.8 GHz), besides adjusting the clock, what else must I tweak to match?
3) After overclocking, do I need to modify RAM or PCIe buses, or can I leave them as is?
Thank you for your attention.
(I’ve taken care of cooling to keep temperatures in check.)
All server CPUs are currently locked. This situation might change later, but for now it's best not to expect any improvements in OCing a server chip. Thank you for your message, tomkis90. You are correct about this. I've already mentioned that:
Pan_Melas:
- The CPU clock ratio works when the clock is unlocked, but I discovered my CPU's clock is locked.
- So I likely won't be able to use this feature.
- The main focus here is whether I can O/C the Xeon and at what level, possibly adjusting other subsystems.
- Any advice would be greatly appreciated.
Hi,
I'm currently experimenting with the same Xeon X5482 3.2GHz (SLANZ) processor (reads as unlocked in my bios with E0 stepping) on my Asus P5Q Pro (P45 chipset) motherboard with Win 7 Pro 64-bit OS. I went to the trouble of flashing the bios with a modded rom which is patched with the Xeon processor microcode. Works fine on defaults, and without changing anything else i can change the CPU host clock from 400 to 425 and achieve 3.4GHz, and it's very stable.
However, i'm trying to push it a little harder than that and had my sights set on 4GHz. After experimentation i think this might be impractical and have decided to aim for either 3.6 or 3.8. At first 3.8 was equally un-achievable as the bios wouldn't even post with a whole host of tweaks. But I was able to get 3.6 by changing the FSB strap clock to 333MHz, CPU clock to 450MHz, and cpu vcore to 1.285v (from searching online i found the default to be 1.25v). I also had to tie down the clock on my memory which in reality is 1066 with 5-5-5-15 timing at 2.1v at default. This i set to something like 1150. I also tried with lower settings like 950 but if anything this was less stable. I should also mention I still have speedstepping enabled so the bios is taking care of the multiplier (auto), depending on workload this is dynamically changing from 6.0 to 8.0 in the OS.
Computer boots into OS, but stress testing with IntelCPUBurnTest showed possible instability. It seems to me that with the right tweaks this could be made to work but as core temps were still within tolerance (below 70c with H60 Corsair water cooling) I decided to take what i'd learn't and push on with 3.8GHz.
I'm currently running vcore at 1.32v (going to lower this if possible but i've seen reports of people taking it up to 1.40 to achieve 4.0GHz on a double cpu Skulltrail mobo), have introduced a pll voltage of 1.15v which i now believe is well below what the board was doing when still in auto. This was enough to get it to boot in to OS for a minute or so before blue screening with memory error. Further increases may yield greater stability, and i'm going to trial the cpu clock skewing to fix the crashing under load.
It's a work in progress but as there is so little information on this subject online, i thought i'd share my initial finding in the hope others might chip in with their success stories!
syphon85 :
Hi,
I'm currently experimenting with the same Xeon X5482 3.2GHz (SLANZ) processor (reads as unlocked in my bios with E0 stepping) on my Asus P5Q Pro (P45 chipset) motherboard with Win 7 Pro 64-bit OS. I went to the trouble of flashing the bios with a modded rom which is patched with the Xeon processor microcode. Works fine on defaults, and without changing anything else i can change the CPU host clock from 400 to 425 and achieve 3.4GHz, and it's very stable.
However, i'm trying to push it a little harder than that and had my sights set on 4GHz. After experimentation i think this might be impractical and have decided to aim for either 3.6 or 3.8. At first 3.8 was equally un-achievable as the bios wouldn't even post with a whole host of tweaks. But I was able to get 3.6 by changing the FSB strap clock to 333MHz, CPU clock to 450MHz, and cpu vcore to 1.285v (from searching online i found the default to be 1.25v). I also had to tie down the clock on my memory which in reality is 1066 with 5-5-5-15 timing at 2.1v at default. This i set to something like 1150. I also tried with lower settings like 950 but if anything this was less stable. I should also mention I still have speedstepping enabled so the bios is taking care of the multiplier (auto), depending on workload this is dynamically changing from 6.0 to 8.0 in the OS.
Computer boots into OS, but stress testing with IntelCPUBurnTest showed possible instability. It seems to me that with the right tweaks this could be made to work but as core temps were still within tolerance (below 70c with H60 Corsair water cooling) I decided to take what i'd learn't and push on with 3.8GHz.
I'm currently running vcore at 1.32v (going to lower this if possible but i've seen reports of people taking it up to 1.40 to achieve 4.0GHz on a double cpu Skulltrail mobo), have introduced a pll voltage of 1.15v which i now believe is well below what the board was doing when still in auto. This was enough to get it to boot in to OS for a minute or so before blue screening with memory error. Further increases may yield greater stability, and i'm going to trial the cpu clock skewing to fix the crashing under load.
It's a work in progress but as there is so little information on this subject online, i thought i'd share my initial finding in the hope others might chip in with their success stories!
Although I know that it has been a long time since my first post on this matter, I am refreshing this thread, quoting your words, in order to ask you if you have achieved what you are initially describing, i.e.your final results with the X5482 O/C (including the settings for voltage and memory speed).
Will you be able to post here your results, in case you have achieved a stable system that works OK?
If yes, please, post as many details as you can, including RAM model, type and settings, voltages, your final choice for cooling method, etc.
This will help me to make my system working correctly because until now I have not been able to O/C my Xeon, been afraid that I may do something wrong and get a disaster at the end.
Thanks in advance.