cpu Significantly degraded CPU?

[Necromancer-_-]

Hi all!

 

So my friend bought a very beefy motherboard (Asus P8P67 Deluxe) to OC his old I5-2400.

 

But it seems like it is significantly degraded due to insanely high temps (99C for hours and hours, or maybe even for months (but with low voltage, so low electro migration) with his before "motherboard", noname brand ...). I could oc his CPU to 4Ghz, and it booted, BUT it was not stable, in CB R15 it scored only 501 since it didnt run on 4Ghz, only around 3.85Ghz or 3.8~Ghz.

The clock was achieved with these settings: 38x105.3 with 1.3V, high LLC, 140% cpu current capability, and power saving options turned off (C1E states, C3, C6, EPU).

I found out that his multiplier was dropping under load (to 36x) so I lowered it and tried to achieve 4ghz with different settings (32x125) but it turns out that the computer wont boot that way above 105~ish BCLK with 32X, or even 36X, it doesnt matter, what it does is it puts the voltage back to 1,103volts (which is obviously not enough for that OC).

 

After a failed OC the MB boots to "Overclocking Failed .... Press F1 to..." so after I enter the bios it shows that 1.103volts, but even if I use tremendously high 1.5Volts just to try it out, it still wouldnt boot (same OC failed screen) and sets the CURRENT volt back to 1.103V (it doesnt change the SET voltage from 1.5 to 1.103V).

 

Tried lowering the ram speed to 1K hz, and to slower timings (from 9 9 9 24 1 to 10 10 10 25 1) but nothing.

 

I made several OC profiles, and it looks damn stable on the 3.5Ghz one he says.

 

So what could be the problem? Dying CPU/Degraded CPU or Dying/Failing/Bugging motherboard? Or possibly bad ram even though it is slowed down (1x8Gb DDR3, dont know the brand now).

 

All answers are appreciated

[Guest]

The constant 99c is why it won't overclock....

 

Why keep trying?

[Somerandomtechyboi]

Tbh just get a 2500k or 2600k and a better cooler then overclock

[Necromancer-_-]

6 hours ago, ShrimpBrime said:

The constant 99c is why it won't overclock....

 

Why keep trying?

You know im just trying to make 101% sure about that.

[Necromancer-_-]

6 hours ago, Somerandomtechyboi said:

Tbh just get a 2500k or 2600k and a better cooler then overclock

The CPU does not overheat now, he has a better cooler on it, max was 61C.

And it doesnt thermal throttle.

 

But of course that doesnt do anything with the degradation now.

[Guest]

6 hours ago, Necromancer-_- said:

You know im just trying to make 101% sure about that.

Well you are now 110% sure you don't know 4ghz would be a great overclock for that chip. 

[Necromancer-_-]

4 minutes ago, ShrimpBrime said:

Well you are now 110% sure you don't know 4ghz would be a great overclock for that chip. 

Well, im not ocing that anymore, he's fine with that 3.5Ghz OC, he said he will buy an I7 later.

[Guest]

2 minutes ago, Necromancer-_- said:

Well, im not ocing that anymore, he's fine with that 3.5Ghz OC, he said he will buy an I7 later.

I think 3.5ghz daily is pretty respectable honestly.

 

Now if 3.5 becomes unstable soon, and a lower clock does stabilize, that would be degradation. 4ghz is just pushing the architecture hard is all.

[Necromancer-_-]

2 hours ago, ShrimpBrime said:

I think 3.5ghz daily is pretty respectable honestly.

 

Now if 3.5 becomes unstable soon, and a lower clock does stabilize, that would be degradation. 4ghz is just pushing the architecture hard is all.

I guess the degradation is around 3.8Ghz, since it didnt drop the clock under load, I didnt see any 4+Ghz oc on that chip anywhere, possibly 4ghz is the top.

[Guest]

4 hours ago, Necromancer-_- said:

I guess the degradation is around 3.8Ghz, since it didnt drop the clock under load, I didnt see any 4+Ghz oc on that chip anywhere, possibly 4ghz is the top.

Lots of 4ghz with that chip with submissions claiming air cooling, but knowing the platform, chilled or big water loop most likely.

 

Tcase max is 72c  keep very far under for stability oc 4ghz.

Junction max 100c = throttle.

[Necromancer-_-]

1 hour ago, ShrimpBrime said:

Lots of 4ghz with that chip with submissions claiming air cooling, but knowing the platform, chilled or big water loop most likely.

 

Tcase max is 72c  keep very far under for stability oc 4ghz.

Junction max 100c = throttle.

His CPU didnt even reach 100C "only" 99C, and it throttled, but im unsure if this was on the die or the cores. well, we cant OC it to 4Ghz now, so temp is not going to be an issue.

He just ordered a better cooler than what he has.

[Guest]

3 hours ago, Necromancer-_- said:

His CPU didnt even reach 100C "only" 99C, and it throttled, but im unsure if this was on the die or the cores. well, we cant OC it to 4Ghz now, so temp is not going to be an issue.

He just ordered a better cooler than what he has.

Right so closer to 1.4v 4ghz range and 1.5v around 5ghz range. Since a 2400 is a pretty low bin chip, 4ghz might be doable, just need the vcore to keep transistors operational.

 

Good luck dudes.

[Necromancer-_-]

9 hours ago, ShrimpBrime said:

Right so closer to 1.4v 4ghz range and 1.5v around 5ghz range. Since a 2400 is a pretty low bin chip, 4ghz might be doable, just need the vcore to keep transistors operational.

 

Good luck dudes.

Thanks!

 

At 4Ghz it doesnt boot with even 1.5V, it just boots into "Failed OC... Press F1....". So probably 3.8 should be doable still. But I was also thinking about what if the motherboard is the one that fails? Like not enough good VRM? It has 16+2 power phases (Which of course doesnt mean that it is superior quality, but it is ASUS it should be OK), and it wasnt even warm, I hope it is not the mobo, cause if he buys another CPU and runs into the same problem then it must be the MB, except if he buys a degraded CPU.

[Guest]

3 hours ago, Necromancer-_- said:

Thanks!

 

At 4Ghz it doesnt boot with even 1.5V, it just boots into "Failed OC... Press F1....". So probably 3.8 should be doable still. But I was also thinking about what if the motherboard is the one that fails? Like not enough good VRM? It has 16+2 power phases (Which of course doesnt mean that it is superior quality, but it is ASUS it should be OK), and it wasnt even warm, I hope it is not the mobo, cause if he buys another CPU and runs into the same problem then it must be the MB, except if he buys a degraded CPU.

It could be a few different variables such as needing to lower the memory divider at higher reference frequencies for example.

[Necromancer-_-]

3 hours ago, ShrimpBrime said:

It could be a few different variables such as needing to lower the memory divider at higher reference frequencies for example.

By "memory divider" what do you mean exactly? I already set the lowest possible ram speed, but it still doesnt boot, or you mean the ram command rate? like 1 or 2? (1T, 2T).

[Guest]

52 minutes ago, Necromancer-_- said:

By "memory divider" what do you mean exactly? I already set the lowest possible ram speed, but it still doesnt boot, or you mean the ram command rate? like 1 or 2? (1T, 2T).

Yes the memory strap. Also called divider. Essentially setting a lower speed.

 

What board is it?

[Necromancer-_-]

46 minutes ago, ShrimpBrime said:

Yes the memory strap. Also called divider. Essentially setting a lower speed.

 

What board is it?

It is in the beggining of the post, but here it is: Asus P8P67 Deluxe.

[Guest]

1 minute ago, Necromancer-_- said:

It is in the beggining of the post, but here it is: Asus P8P67 Deluxe.

I saw after I had asked haha, sorry.

 

Have to see what better cooling does. My 2500K hit 5ghz, but on a custom water loop. I can't remember if I had the chiller hooked up or not honestly. That was a while ago.

[Necromancer-_-]

12 minutes ago, ShrimpBrime said:

I saw after I had asked haha, sorry.

 

Have to see what better cooling does. My 2500K hit 5ghz, but on a custom water loop. I can't remember if I had the chiller hooked up or not honestly. That was a while ago.

No problem :D.

 

Since I live like 35-40kms away from him, I wont really be able to try it out it with the new cooler he ordered. Probably just months later. But I'm curious about that, maybe it really does OC better with a better cooler, because the one that he has right now, is basically a home made cooler (some beefy heatsink + an intel stock fan + zip ties), but it does good for a home made cooler.

[Necromancer-_-]

6 minutes ago, Necromancer-_- said:

No problem :D.

 

Since I live like 35-40kms away from him, I wont really be able to try it out it with the new cooler he ordered. Probably just months later. But I'm curious about that, maybe it really does OC better with a better cooler, because the one that he has right now, is basically a home made cooler (some beefy heatsink + an intel stock fan + zip ties), but it does good for a home made cooler.

Right now I'm using an older CPU mainly for OC (I7-920 2.67Ghz D0 stepping), and with a good motherboard (Asus P6T SE) I could overclock it to 3.7Ghz @ just 1.10625V, but for 4Ghz it needs 1.25V. With a Deepcool Gammaxx 300 cooler. I'm surprised how good this overclocks knowing the fact that it came out in 2008.

[thewelshbrummie]

Surprised there's this number of comments without anyone mentioning that the 2400 is not a K CPU and therefore can't be overclocked in the usual manner. You're only option is changing the BCLK. That said:

On 8/26/2021 at 5:50 PM, Necromancer-_- said:

I found out that his multiplier was dropping under load (to 36x) so I lowered it and tried to achieve 4ghz with different settings (32x125) but it turns out that the computer wont boot that way above 105~ish BCLK with 32X, or even 36X, it doesnt matter, what it does is it puts the voltage back to 1,103volts (which is obviously not enough for that OC).

You can't raise the BCLK that high (to 125) and increasing it to overclock at all is exteremely risky as all sorts of things are tied to that, including the SATA controller. It's just not worth messing with as system instability is pretty much guaranteed, especially if you go over 105MHz (and I think data loss is a possibility if you're using a SATA drive as they expect a 100MHZ BCLK so you risk bricking the drive). That's one of the first things I remember reading about Sandy Bridge when it launched 10 years ago, which is the reason Intel do the upsell to -K CPUs, stick with a BCLK of 100 and rise the multiplier to whatever you can get to work.

 

FYI 1st Gen Core i7 (Nehalem and Westmere), like your 920, was essentially an early HEDT platform - so despite not being a K CPU, overclocking is an option as the multiplier is unlocked. Looking at the CPU lists the only CPU on 1st Gen Intel that was overclockable on the mainstream Lynnfield or Clarkdale platforms was the i7-875K - everything else was multiplier locked.

 

From memory, a BCLK of 105 is the absolute maximum you can go - 103 is fine but the benefits will be so minimal that it just isn't worth it.

 

And CPUs don't degrade - they either run within spec or flat out stop working. My 12 year old Core 2 Quad Q9650 is just as fast as it was new - it's just not fast enough anymore for new games, but it certainly hasn't lost speed or performance relative to what it could do when new.

 

As others have said, best option is a 2500K or 2600K - or if the money is available, a new build.

[Guest]

34 minutes ago, Necromancer-_- said:

Right now I'm using an older CPU mainly for OC (I7-920 2.67Ghz D0 stepping), and with a good motherboard (Asus P6T SE) I could overclock it to 3.7Ghz @ just 1.10625V, but for 4Ghz it needs 1.25V. With a Deepcool Gammaxx 300 cooler. I'm surprised how good this overclocks knowing the fact that it came out in 2008.

I have a 930 a 980 and a 980x with some xeon chips and a gigabyte board. 

 

Rocked out some decent overclocks, but with an 8c water delta. 

 

Fun platform. 

[Necromancer-_-]

2 hours ago, thewelshbrummie said:

Surprised there's this number of comments without anyone mentioning that the 2400 is not a K CPU and therefore can't be overclocked in the usual manner. You're only option is changing the BCLK. That said:

You can't raise the BCLK that high (to 125) and increasing it to overclock at all is exteremely risky as all sorts of things are tied to that, including the SATA controller. It's just not worth messing with as system instability is pretty much guaranteed, especially if you go over 105MHz (and I think data loss is a possibility if you're using a SATA drive as they expect a 100MHZ BCLK so you risk bricking the drive). That's one of the first things I remember reading about Sandy Bridge when it launched 10 years ago, which is the reason Intel do the upsell to -K CPUs, stick with a BCLK of 100 and rise the multiplier to whatever you can get to work.

 

FYI 1st Gen Core i7 (Nehalem and Westmere), like your 920, was essentially an early HEDT platform - so despite not being a K CPU, overclocking is an option as the multiplier is unlocked. Looking at the CPU lists the only CPU on 1st Gen Intel that was overclockable on the mainstream Lynnfield or Clarkdale platforms was the i7-875K - everything else was multiplier locked.

 

From memory, a BCLK of 105 is the absolute maximum you can go - 103 is fine but the benefits will be so minimal that it just isn't worth it.

 

And CPUs don't degrade - they either run within spec or flat out stop working. My 12 year old Core 2 Quad Q9650 is just as fast as it was new - it's just not fast enough anymore for new games, but it certainly hasn't lost speed or performance relative to what it could do when new.

 

As others have said, best option is a 2500K or 2600K - or if the money is available, a new build.

Yes no one mentioned it, and I was curious even if I couldnt raise it to 40+, how could I change it anyway if it is locked?

 

I see, I knew BCLK messes around with other PC parts other than the CPU, but I didnt know it includes even the SATA drives, thanks to pointing this out!

If 105~ is kinda the "max" it can go, then how some people can go over like 180+? I mean I know my old I7 is different architecture, but it is currently running at 185Mhz BCLK which is way over 133, and it has no problems that I can see. Though it is unstable at 21X multiplier, 20X is which it runs on now.

 

By not degrading you mean that his CPU can run on 99C until it dies without any visible/noticable degradation or performance loss? So if an XY CPU runs on insanely high temps for months or for years still runs the same (same performance, even overclockabilty) compared to a same new CPU (EXACTLY the same CPU, forget silicon lottery now)?

 

Yes it would be, he just bought the Motherboard, so he wont invest in a new build, he said he will get an I7 later (probably a 2600K) but not now.

[Necromancer-_-]

31 minutes ago, ShrimpBrime said:

I have a 930 a 980 and a 980x with some xeon chips and a gigabyte board. 

 

Rocked out some decent overclocks, but with an 8c water delta. 

 

Fun platform. 

Wow, I also want to try out some xeon chips (especially the X5690) bit later. Those 980/980X chips are pretty expensive even now.

 

LN2 then?

[Guest]

1 hour ago, Necromancer-_- said:

Wow, I also want to try out some xeon chips (especially the X5690) bit later. Those 980/980X chips are pretty expensive even now.

 

LN2 then?

No ln2, well none on this platform. I don't have a spare board, otherwise I probably would. Think the 980 maxed out at 5.2ghz or thereabouts. https://hwbot.org/submission/4687323_shrimpbrime_cpu_frequency_core_i7_980_5251.84_mhz