i5-750 vs P55 vs water Currently @4.4ghz!

[Kenmitch]

Edited Part

My take on this whole paranoid factor is in post #24 would take to much space up here

For the ultra paranoid guys I decided to play around within the paranoid zone I like to call it.

The chip was stable @4ghz with a vcore of 1.325 and cpu vtt of 1.18

Working on 4.1ghz and looks like it'll be vcore of 1.360 and cpu vtt of 1.21-1.22

For those whom aren't as paranoid the chip would stabalize with a cpu vtt of 1.38 @4.3ghz

I'm thinking tonight I'll try 4.4ghz again and see at what point it will stabalize at.

After a shakey start on Sunday night I got a chance to start some real testing on my chip. I made the mistake of updating my bios to the current one on Sunday night. It turns out that the current bios for my board is giving alot of us tweakers a rough time

After I downgraded bios this AM I started testing my chip out. What a diff a bios makes at least in my case anyways.

Currently I have the following still enabled in bios. C1E,speedstep,turbo mode. Currently the board is running with vdroop. Guess maybe I should try without vdroop also.

I'm sitting at 4.3ghz (205x21)at the moment with the following settings....In the images

Now the temps are starting to get kinda high for being on water.Guess I could maybe speed up my fans

The temps are from Intel Burn Test. OCCT temps are about 5-7 or so *C cooler for each core.

My main concern is the voltages. Vcore is kinda getting scary but more concerned about the other voltages as it seems like they may need some tweaking. Any room to play on them?

Most likely I'll back her down as I'd kinda like to play around with the 2 core 24x turbo as windows 7 seems to be pretty good at shifting the load from core to core to keep the chip cooler anyways.

Kinda on a tweaking break at the moment untill my daughter goes to sleep tonight. I figured I'd ask for some input from others who have had more knowledge in the area.

Thanks,
Ken

 

[The Sauce]

I have read that 1.55v is max. I got as high as 1.425v and hit 4.4GHz but started getting blocked out by high temps. I personally am not comfortable running higher than 1.35v for 24/7 but 1.40 is probably ok on water - at least according to Intel specs.

If I remember VTT 1.4v is kinda high. I would probably go to 1.2v.

 

[n7]

Very nice clocks, but that's rather high VTT for 24/7.

I'd suggest lowering that if possible.

 

[RussianSensation]

I would create an excel chart with CPU speed, CPU Vcore and QPI/Vtt and corresponding CPU load temperatures. You should be able to figure out the sweet spot. For example, in my system going from 3.9 to 4.0ghz, takes CPU temperatures another 7-8*C, requires CPU voltage of 1.36V+ and QPI/Vtt of 1.41V from 1.25V. The difference between 4.3ghz and 4.1ghz is <5&#37;, but if you can lower your CPU voltages < 1.40 and QPI/VTT closer to 1.21V maximum limit, I would sacrifice the 5% performance difference. Just my 2 cents.

 

[Kenmitch]

Very nice clocks, but that's rather high VTT for 24/7.

I'd suggest lowering that if possible.

Guess I got lucky again as my goal was a 4ghz plus chip

I'm going to play around with it later tonight when my daughter goes to sleep.

The cpu vtt is on auto guess I'll try to lower it and see at whats the lowest it'll stay stable at. This was one voltage adjustment I was unsure of as most of the upper end oc results I've seen so far have a similar cpu vtt some of them way higher.

I'm just kinda trying to get a feel for what the chip will do.

 

[Kenmitch]

I would create an excel chart with CPU speed, CPU Vcore and QPI/Vtt and corresponding CPU load temperatures. You should be able to figure out the sweet spot. For example, in my system going from 3.9 to 4.0ghz, takes CPU temperatures another 7-8*C, requires CPU voltage of 1.36V+ and QPI/Vtt of 1.41V from 1.25V. The difference between 4.3ghz and 4.1ghz is <5%, but if you can lower your CPU voltages < 1.40 and QPI/VTT closer to 1.21V maximum limit, I would sacrifice the 5% performance difference. Just my 2 cents.

You know how it is. Man gets a new toy and has to play with it.

That sweet spot is gonna be around 4-4.1ghz on my chip I already figured it out.

Now I'm just trying to find the max the chip will do. As I'm thinking about just getting the 2 core 24x multiplier fired up.

Looks like the chip has a shot at it working....Just need to do some more testing and find out.

If all goes as planned this chip should wind up faster or equal to my x Q9550 @ 4ghz under 4 core load but should smoke it under 2 core load

Anyways if it doesn't work out that way then at least I get to play around and kill some time. Worst case senario I'll have a i5-750 at 4ghz plus!

 

[The Sauce]

Guess I got lucky again as my goal was a 4ghz plus chip

Seems the new batch of 750s are doing 4.0-4.4GHz on air pretty easily.

 

[Kenmitch]

Seems the new batch of 750s are doing 4.0-4.4GHz on air pretty easily.

We've got very similar systems

I had this one up to 4.4ghz last night....But am backing her down now as I'm kinda looking for a compromise of speed vs heat vs voltage at the moment.

Higher stable bclk is better performance whise vs multiplier? Same overall speed

Kinda trying to get the feel of this new platform

 

[n7]

Higher stable bclk is better performance whise vs multiplier? Same overall speed

Kinda trying to get the feel of this new platform

Yes, but you'll only see the benefit in synthetic benches, & high BCLK tends to require more VTT as well.

 

[Kenmitch]

Yes, but you'll only see the benefit in synthetic benches, & high BCLK tends to require more VTT as well.

That is kinda what I was thinking. To be considered a high BCLK it would be above 180? 200?

Thanks,
Ken

 

[MrK6]

That is kinda what I was thinking. To be considered a high BCLK it would be above 180? 200?

Thanks,
Ken

Higher than stock is high BCLK . Performance scales with speed, it's just that you'll only see the difference in synthetic benchmarks, since the BCLK clock isn't the limiting factor in performance in applications. However, BCLK does limit your overclock since it requires much more voltage at ~195+. Therefore, it's best to use your highest available multiplier and go for the fastest CPU speed while remaining within Intel's specs.

 

[Kenmitch]

Higher than stock is high BCLK . Performance scales with speed, it's just that you'll only see the difference in synthetic benchmarks, since the BCLK clock isn't the limiting factor in performance in applications. However, BCLK does limit your overclock since it requires much more voltage at ~195+. Therefore, it's best to use your highest available multiplier and go for the fastest CPU speed while remaining within Intel's specs.

Intel's specs ???

cpu vtt as close to 1.25 as possible and maybe 1.35 as a max with the lower the better ???
cpu vcore anything under 1.45 on water should be OK....No vid on i's ???

Still trying to figure out how to lock down the voltages on my MB....Kinda seems like either the board or the chip over rides my settings in bios ???

Thanks,
Ken

 

[MrK6]

Intel's specs ???

cpu vtt as close to 1.25 as possible and maybe 1.35 as a max with the lower the better ???
cpu vcore anything under 1.45 on water should be OK....No vid on i's ???

Still trying to figure out how to lock down the voltages on my MB....Kinda seems like either the board or the chip over rides my settings in bios ???

Thanks,
Ken

According to Intel, maximum voltages are 1.21V VTT, 1.55V Vcore, 1.65V Vdimm, and 1.98V CPU PLL (don't quote me on that last one, but I think I'm correct). The low VTT really cripples the ability to crank BCLK clocks, but my attitude is that they wouldn't specify it if it didn't mean something. Hell, I'm actually of the opinion that's why there was that commotion at socket 1156's launch where some overclockers were completely melting their sockets; my guess is they were using the same VTT's of 1.35V+ they had been using with Bloomfield chips and it was too much. Anywho, Vcore shouldn't be a problem, you'll most likely reach some thermal limit before you hurt the chip assuming Intel's 1.55V spec is truly safe. As far as the voltages go, I haven't used your motherboard but make sure you're using the latest BIOS and then set as many voltages as necessary manually (not Auto, etc.).

 

[Kenmitch]

As far as the voltages go, I haven't used your motherboard but make sure you're using the latest BIOS and then set as many voltages as necessary manually (not Auto, etc.).

Well Sunday when I gotta chance to start tweaking the system I updated to the latest bios. The latest bios is buggy as hell and would pretty much not work at all for me. I kept getting stop errors. I was forced to backpeddle my bios and went back 2 versions to be safe as that was what I thought the board shipped with. But in the end I wound up updating it by one....This one seems to be working pretty good as far as overclocking anyways. I just gotta figure out how to lock down the voltages now.

I'm currently sitting at 4.1ghz (205x20) and she is pretty stable so far. I'm kinda using the evga e-leet program to drop the voltages on the MB to find the lowest stable voltages the chip will run at and just writing them down for now.

loaded vcore is 1.35 and the cpu vtt is down to 1.26 the temps are in the 50's

Guess I could save this profile and maybe try upping the multiplier and lowering the BCLK and redo the voltages to see if they come out much diff.

Thanks

 

[astrosfan315]

wow 205 FSB. What about increasing the multiplier and decreasing FSB? Or did you just reach a point where 205 x21 or 195x22 equaled the same? I just started messing around with my i5-750 last night and need to get used to the BIOS more than anything. Just curious about the multiplier vs pure FSB.

 

[Kenmitch]

wow 205 FSB. What about increasing the multiplier and decreasing FSB? Or did you just reach a point where 205 x21 or 195x22 equaled the same? I just started messing around with my i5-750 last night and need to get used to the BIOS more than anything. Just curious about the multiplier vs pure FSB.

Stock multiplier is 20x with turbo enabled it's 21x.The 22x 23x 24x multipliers are only active under 1-2 core loads.

 

[Kenmitch]

Started over @ 4.1ghz and have stabalized up to 4.3ghz so far

Currently working on 4.4ghz!

I think I crossed the line into the extreme zone!

No suicide runs for me yet.

Board seems pretty damn stable so far but I'm not sure bow much more BLCK she has left in her.

 

[EarthwormJim]

Started over @ 4.1ghz and have stabalized up to 4.3ghz so far

Currently working on 4.4ghz!

I think I crossed the line into the extreme zone!

No suicide runs for me yet.

Board seems pretty damn stable so far but I'm not sure bow much more BLCK she has left in her.

I'd really be concerned with the Vtt voltages you're probably having to use to clock that high (Bclk over 200).

Intel was pretty specific with those voltages, and it's not like they were overly conservative with the rest of the voltages on the chip. They're ok with 1.55v on the vcore.

 

[Kenmitch]

I'd really be concerned with the Vtt voltages you're probably having to use to clock that high (Bclk over 200).

Intel was pretty specific with those voltages, and it's not like they were overly conservative with the rest of the voltages on the chip. They're ok with 1.55v on the vcore.

Well you kinda take all the fun out of it when you put it that way

You think my voltages are kinda on the high side then maybe you should take a look at this thread and see what one guy is willing to do just to make his dud chip fly.

http://www.evga.com/forums/tm.aspx?m=132581

 

[EarthwormJim]

Well you kinda take all the fun out of it when you put it that way

You think my voltages are kinda on the high side then maybe you should take a look at this thread and see what one guy is willing to do just to make his dud chip fly.

http://www.evga.com/forums/tm.aspx?m=132581

I've read the evga forum posts about Vtt when I was overclocking my Lynnfield. I take their advice with a grain of salt though, they're basically justifying the voltages based on Bloomfield's specs despite the IMC being of a different design.

Intel was frightening specific with their Vtt voltage, and very relaxed with the other voltage specs. That makes me nervous.

Could also be marketing, maybe they want to scare overclocking enthusiasts away from Lynnfield and towards Bloomfield. x58 is supposed to be the enthusiast platform.

 

[Munky]

I have a feeling a bunch of users running 1.35v vtt on lynnfield will be waking up to a dead cpu sooner than later.

 

[n7]

I have a feeling a bunch of users running 1.35v vtt on lynnfield will be waking up to a dead cpu sooner than later.

Honestly, i really doubt it.

We heard the same thing w/ high VTT on our s775 quads, especially the 45nm ones, & how many people have you heard of killing them w/ high VTT?

Yeah, exactly.

I always advise erring on the side of caution, but that said, i don't think it's very likely in the first place.

Kenmitch, you've got a really nice CPU
I could only do around 4.15 GHz LinX stable, though i suspect it had something to do w/ 8 GB.
I could do 4.4 GHz for screenshots, but it certainly wasn't anywhere close to stable :hmm:

 

[MrK6]

Tough to say. By design socket 1156 has fewer power pins going to the CPU anyway. This might have something to do with the revamped voltage specs and the fact that we saw so many damage 1156 sockets at release, or it could have nothing to do with it. I'm not an engineer so I don't know. I'm still not going to be the one to test it out though .

 

[Kenmitch]

Honestly, i really doubt it.

We heard the same thing w/ high VTT on our s775 quads, especially the 45nm ones, & how many people have you heard of killing them w/ high VTT?

Yeah, exactly.

I always advise erring on the side of caution, but that said, i don't think it's very likely in the first place.

Kenmitch, you've got a really nice CPU
I could only do around 4.15 GHz LinX stable, though i suspect it had something to do w/ 8 GB.
I could do 4.4 GHz for screenshots, but it certainly wasn't anywhere close to stable :hmm:

My theory on this issue would be as follows. Each individual chip will determine when it's at the point of failure. When it's had enought it'll let you know.

If you look at all the great bang for buck intel chips from the past you will notice that this is kinda the norm. If your lucky and you get one of the great ones the chip will scale nicely and it will let you know when it's had enough. When you start to pump massive amounts of juice in the chip to try and make it a great one that is when you will have a keychain

I'll use and e5200 for example as that is when I jumped back to intel. If your lucky and get a good one you can reach 4ghz plus on a $60 motherboard as long as it's got a overclocking friendly bios. If your unlucky and don't get one of the good ones your just not gonna make the 4ghz mark even if you put the baby in the top of the line latest greates $300 motherboard. Even if you keep pumping more and more juice into the chip it won't help.

This has been true pretty much for the entire intel line of cpus so why would it all of a sudden just change?

Is this cpu vtt spec of 1.21 max conservative?
Was it pulled out of the intel's arse to deter overclocking?
Why does intel market the i5-750 @2.66ghz when it runs @2.79ghz unless you disable turbo?
Is there much of a diff of a cpu vtt at 1.21 vs 1.25 vs 1.28 and so on?

Back to my chip. Last night I was kinda feeling paranoid for some unkown reason so I decided to back her down to 4ghz and play around for awhile. It was stable with a vcore around 1.325 with a cpu vtt of 1.18 which falls into the ultra paranoid overclocking category

Now does bumping the voltage a couple of ticks and bumping the cpu vtt a tick or two put the chip in the danger zone to make the next jump mean the chip will fry....I doubt it

I'm still stuck on the theory that nothing has changed from the intel chips of the past to the present day intel chips.

The chip will determine the point of failure. Not the laser etching or the imbedded chip ID

 

[Kenmitch]

I'm still not going to be the one to test it out though .

Nobody asked you to test it out.

If you just do some googling for i5 750 or i7 860 you will find many guides or results with people who are testing it out for you.