Should I upgrade CPU (and Mobo)?

GamingDaemon

Senior member
Apr 28, 2006
474
7
76
HI Folks,

I'm mostly a gamer. Do a little video editing, writing, programming, etc. But mostly I game on my PC (see my sig below).

So should I upgrade to an Intel i3570k and to the 1155 socket? The total cost, including a new Fractal case would be around $515, which is do-able.

Now, I should say, all my games play quite nicely. And I have zero problems with my machine.

But I am in this forum because I am a nerd, a tech geek, a CPU-phile, and I want to help out the economy.

So what would my advantages be if I upgraded? (yes, this may end up being more of a philosophical than technical thread)

Thanks in advance...
 

BonzaiDuck

Lifer
Jun 30, 2004
16,664
2,039
126
HI Folks,

I'm mostly a gamer. Do a little video editing, writing, programming, etc. But mostly I game on my PC (see my sig below).

So should I upgrade to an Intel i3570k and to the 1155 socket? The total cost, including a new Fractal case would be around $515, which is do-able.

Now, I should say, all my games play quite nicely. And I have zero problems with my machine.

But I am in this forum because I am a nerd, a tech geek, a CPU-phile, and I want to help out the economy.

So what would my advantages be if I upgraded? (yes, this may end up being more of a philosophical than technical thread)

Thanks in advance...

Only you can answer that question. Yes, there are performance advantages to SB and IB over socket-1366. For me, I seem to hop-scotch a chip generation between builds. It all seems to be about budget, need and want.

If you want to over-clock, the IB processor only seems to show its stuff if you mitigate its thermal limitations through de-lidding -- which has some risk and means doing some work. I'm happy with my SB 2600K @ 4.6Ghz (Turbo). I've put aside thoughts of building an Ivy Bridge machine for now.
 

GamingDaemon

Senior member
Apr 28, 2006
474
7
76
Only you can answer that question. Yes, there are performance advantages to SB and IB over socket-1366. For me, I seem to hop-scotch a chip generation between builds. It all seems to be about budget, need and want.

If you want to over-clock, the IB processor only seems to show its stuff if you mitigate its thermal limitations through de-lidding -- which has some risk and means doing some work. I'm happy with my SB 2600K @ 4.6Ghz (Turbo). I've put aside thoughts of building an Ivy Bridge machine for now.

I definitely want to overclock, but delidding makes me a bit nervous. Should I wait for Haswell?
 

BonzaiDuck

Lifer
Jun 30, 2004
16,664
2,039
126
I definitely want to overclock, but delidding makes me a bit nervous. Should I wait for Haswell?

You know -- at this point, I can't say. The die-size for Haswell will be again smaller than for IB. I can't remember if it had been discussed here on the forums, or we don't know what Intel will do with their Ivy Bridge departure from using fluxless silver-indium solder between the cap and the processor.

Put it another way -- it's likely to be "just as fast" while using less power, and there would be an improvement in features (like the iGPU).

You CAN overclock Ivy Bridge; some here seem to have pegged expectations at 4.4 to 4.5 Ghz in Turbo Mode. Others who de-lidded the processor-cap and made modifications for a "direct-die" waterblock or heatsink application seemed to indicate that 4.9 is possible, and 4.6 to 4.8 is reasonable. Their temperatures were considerably lower for the removal of the IHS-cap and application of various TIM replacements. But at 4.9 Ghz, the voltage -- around 1.4V -- seems likely to shorten processor life. I think the cautious limit was around 1.30V, and the tests showed reasonable temperatures with speeds around 4.7 or 4.8 at that voltage. And really -- more like 4.7 Ghz, at that . . .
 

GamingDaemon

Senior member
Apr 28, 2006
474
7
76
You know -- at this point, I can't say. The die-size for Haswell will be again smaller than for IB. I can't remember if it had been discussed here on the forums, or we don't know what Intel will do with their Ivy Bridge departure from using fluxless silver-indium solder between the cap and the processor.

Put it another way -- it's likely to be "just as fast" while using less power, and there would be an improvement in features (like the iGPU).

You CAN overclock Ivy Bridge; some here seem to have pegged expectations at 4.4 to 4.5 Ghz in Turbo Mode. Others who de-lidded the processor-cap and made modifications for a "direct-die" waterblock or heatsink application seemed to indicate that 4.9 is possible, and 4.6 to 4.8 is reasonable. Their temperatures were considerably lower for the removal of the IHS-cap and application of various TIM replacements. But at 4.9 Ghz, the voltage -- around 1.4V -- seems likely to shorten processor life. I think the cautious limit was around 1.30V, and the tests showed reasonable temperatures with speeds around 4.7 or 4.8 at that voltage. And really -- more like 4.7 Ghz, at that . . .

Well, 4.7GHz is far more than what I would be interested in to obtain. I'd be ecstatic with a 4.0GHz overclock on the 3570k, but I'm not sure what that would do to Turbo Mode. Would I need to disable it?
 

BonzaiDuck

Lifer
Jun 30, 2004
16,664
2,039
126
Well, 4.7GHz is far more than what I would be interested in to obtain. I'd be ecstatic with a 4.0GHz overclock on the 3570k, but I'm not sure what that would do to Turbo Mode. Would I need to disable it?

The processor works at stock settings to run at a base speed at idle and a turbo speed on demand. My i7-2600K for instance has a base-clock of 3.4Ghz and a turbo-speed of 3.8.

Throw in the EIST "Intel Speedstep" energy-saving feature, which downclocks the processor at idle. Thus, your idle speed would then drop from 3.4 (on my CPU) to about 1.6Ghz. So my CPU rides between 1.6 and 4.6Ghz -- over-clocked in "turbo" mode. The voltage rides between 1.008V to an unloaded turbo-mode voltage of about 1.35+V -- which is a reasonably good result with the Sandy Bridge core.

With the socket-1155, you can over-clock while leaving EIST, C1E and other energy-saving features "Enabled."

Once was the case that we just turned those things off, and fixed the processor voltage to run idle or load at the desired speed.

But there's no reason to do that anymore. There's less stress on the processor, you save money on your electric bill. And you get the full, stable desired speed for overclocking the "Turbo" mode.

Lemme review the specs on the i5 you have there . . . Yeah -- that's the same as the 2600K Sandy. 3.4 to 3.8. I'm pretty sure people were able to get to 4.4Ghz with some amount of ease without de-lidding the processor. On those processors, there will be some variability in the thermal solution, because of variability in manufacture in application of Intel's own TIM between the IHS-cap and the processor die. Some will run warmer, others less.

Also, with that expectation of maybe 4.4, the maximum turbo voltage should be sufficiently below 1.30V -- not sure exactly what it would be -- maybe 1.25. IDontCare's OC'ing experiments with the i7-3770K had it pushing 1.298V at about 4.7Ghz.

Whatever the minimum stable turbo voltage you get, the CPU will spend a lot less time at that voltage than if you simply fix the clock-speed and voltage for the maximum that you want.
 

GamingDaemon

Senior member
Apr 28, 2006
474
7
76
The processor works at stock settings to run at a base speed at idle and a turbo speed on demand. My i7-2600K for instance has a base-clock of 3.4Ghz and a turbo-speed of 3.8.

Throw in the EIST "Intel Speedstep" energy-saving feature, which downclocks the processor at idle. Thus, your idle speed would then drop from 3.4 (on my CPU) to about 1.6Ghz. So my CPU rides between 1.6 and 4.6Ghz -- over-clocked in "turbo" mode. The voltage rides between 1.008V to an unloaded turbo-mode voltage of about 1.35+V -- which is a reasonably good result with the Sandy Bridge core.

With the socket-1155, you can over-clock while leaving EIST, C1E and other energy-saving features "Enabled."

Once was the case that we just turned those things off, and fixed the processor voltage to run idle or load at the desired speed.

But there's no reason to do that anymore. There's less stress on the processor, you save money on your electric bill. And you get the full, stable desired speed for overclocking the "Turbo" mode.

Lemme review the specs on the i5 you have there . . . Yeah -- that's the same as the 2600K Sandy. 3.4 to 3.8. I'm pretty sure people were able to get to 4.4Ghz with some amount of ease without de-lidding the processor. On those processors, there will be some variability in the thermal solution, because of variability in manufacture in application of Intel's own TIM between the IHS-cap and the processor die. Some will run warmer, others less.

Also, with that expectation of maybe 4.4, the maximum turbo voltage should be sufficiently below 1.30V -- not sure exactly what it would be -- maybe 1.25. IDontCare's OC'ing experiments with the i7-3770K had it pushing 1.298V at about 4.7Ghz.

Whatever the minimum stable turbo voltage you get, the CPU will spend a lot less time at that voltage than if you simply fix the clock-speed and voltage for the maximum that you want.

I didn't know that the 1155 socket allowed you to overclock with EIST and C1E enabled. That's a reason to buy right now! :) Great explanation.


Thank you. Reading this now.