Max OC for E8400, Q9450

[VirtualLarry]

Reading a few posts here, it seems like not everyone is able to reach 4.0 Ghz easy with an E8400. Taking that into account, and how high a Q9450 will theoretically OC (I'm not sure, but I've seen 3.5Ghz mentioned), then which CPU would be a better buy?

 

[The-Noid]

8x450 = 3600 will be pretty standard.

 

[Tempered81]

9x450 = 4050 that will be standard on an E8400

It's half the price & half the cores. You decide from there.

q9450 is $373 @ bestbuy
http://www.bestbuybusiness.com...95&websrc=FRBB10837495

e8400 is @189 @ microcenter


This is a good question. In two or three years, doom 4, farcry 3, halo 4, crysis 2, call of duty 5, and fear 3 might all be multithreaded.

an old q6600 at that point might be faster than an old e8400 when that time comes.

Who knows what will happen. When you go to fire up doom 3, farcry 2, halo 3, crysis 1, and call of duty 4, the dual core will be faster.

Its an enigma!

 

[The-Noid]

q9450 is 8x not 9x afaik. That will limit this chip.

 

[Tempered81]

yah, it's 8x, the e8400 is the 9x

You think the avg Overclock will see 450mhz fsb on the 45nm quads?

 

[harpoon84]

Originally posted by: jaredpace
yah, it's 8x, the e8400 is the 9x

You think the avg Overclock will see 450mhz fsb on the 45nm quads?

On a decent P35 or X38 board, yes. In fact, it should go slightly higher, the ceiling is around the 470FSB mark from what I've seen.

So realistically, around 3.75GHz or so would be the limit on Q9450s.

 

[tenax]

i'm able at this point to do either 7.5 x 450=3375 stable or 8 x 427=3416 stable on my penryn 9450 (B1 es processor) on an abit ip35 pro. that's with mushkin 6400 memory (2 gigs) and when i try to go higher than that in either case, i get a C1 post error code which should indicate memory. i have 8500 on order and should receive monday. i'll report back then if i can go any higher.

 

[graysky]

@tenax - you have a q9450? Any chance I can get your to run my x264 benchmark with that system @ 8x333 and your 7.5x450?

Here is the link to the benchmark.

 

[tenax]

sure...i'll give it a whirl

 

[aigomorla]

well let me just comment about the E8400 because i too now have one.

4.0ghz is probably the upper limit on air to most people.

the voltage requirement i need for 4.0 -> 4.3 is rediculously high for the 300mhz.

im currently pushing her at 4.3 on 1.57Vcore(NOT SAFE) for a stable 7 hour prime. Its under water tho with 4.5mm d-tek accelerator nozzles.

4.0ghz on my chip to hold 7 hour prime is @ 1.4Vcore. ive seen as low other tho as low as 1.36Vcore.

the sweet oc spot voltage/clock ratio on my chip will end at around ~4.1ghz. From there on the voltage increase looks almost exponential.

The highest votlage i can push this chip on my setup with it dying right away is 1.6V 4.5ghz is impossible at 1.6V even with my setup. :T

Oh the board will do 500fsb ive tried 8x500.
Im having fun with my DFI X38-T2R :T



The Q9450... i have no comment on yet... let me have a bit more fun with this E8400 on my bench setup.

Originally posted by: harpoon84

Originally posted by: jaredpace
yah, it's 8x, the e8400 is the 9x

You think the avg Overclock will see 450mhz fsb on the 45nm quads?

On a decent P35 or X38 board, yes. In fact, it should go slightly higher, the ceiling is around the 470FSB mark from what I've seen.

So realistically, around 3.75GHz or so would be the limit on Q9450s.

Assuming you could keep the temps in check:
Actually im thinkn a GOOD G0 Q6700 clocked @ 4.0ghz would give a Q9450 clocked @ 3.6-3.8 a run for its money. :T

xbit labs showed that quads will have a very hard time passing 450fsb. Dont even think about 500fsb. And 450+ will require a GREAT board to keep and hold for long term.

 

[graysky]

Originally posted by: tenax
sure...i'll give it a whirl

Cool man, thanks. I'm also interested in the vcore and your core temps as well as your MB temps @ 8x427 for example. I'm seriously thinking about getting a q9450 and a Asus P5K in March.

Thanks!

 

[harpoon84]

Originally posted by: aigomorla
Assuming you could keep the temps in check:
Actually im thinkn a GOOD G0 Q6700 clocked @ 4.0ghz would give a Q9450 clocked @ 3.6-3.8 a run for its money. :T

xbit labs showed that quads will have a very hard time passing 450fsb. Dont even think about 500fsb. And 450+ will require a GREAT board to keep and hold for long term.

It certainly would, but how much power would a hypothetical 4GHz Q6700 pull? Keep in mind that a Q9450 should hit it's FSB induced limit quite comfortably, at most only a small Vcore bump should be needed for 3.6 - 3.7GHz.

Even at 3.6GHz, a Q6600 is already pulling an extra 70W under load compared to a Q9300 @ 3.5GHz. I'm afraid trying to milk that final 400MHz from a 65nm Kentsfield would be akin to you trying to squeeze that final 300MHz from your E8400 - the volts/power would just skyrocket and make it unfeasible to achieve with standard air cooling.

 

[JAG87]

Originally posted by: aigomorla
well let me just comment about the E8400 because i too now have one.

4.0ghz is probably the upper limit on air to most people.

the voltage requirement i need for 4.0 -> 4.3 is rediculously high for the 300mhz.

im currently pushing her at 4.3 on 1.57Vcore(NOT SAFE) for a stable 7 hour prime. Its under water tho with 4.5mm d-tek accelerator nozzles.

4.0ghz on my chip to hold 7 hour prime is @ 1.4Vcore. ive seen as low other tho as low as 1.36Vcore.

the sweet oc spot voltage/clock ratio on my chip will end at around ~4.1ghz. From there on the voltage increase looks almost exponential.

The highest votlage i can push this chip on my setup with it dying right away is 1.6V 4.5ghz is impossible at 1.6V even with my setup. :T

Oh the board will do 500fsb ive tried 8x500.
Im having fun with my DFI X38-T2R :T



The Q9450... i have no comment on yet... let me have a bit more fun with this E8400 on my bench setup.

Originally posted by: harpoon84

Originally posted by: jaredpace
yah, it's 8x, the e8400 is the 9x

You think the avg Overclock will see 450mhz fsb on the 45nm quads?

On a decent P35 or X38 board, yes. In fact, it should go slightly higher, the ceiling is around the 470FSB mark from what I've seen.

So realistically, around 3.75GHz or so would be the limit on Q9450s.

Assuming you could keep the temps in check:
Actually im thinkn a GOOD G0 Q6700 clocked @ 4.0ghz would give a Q9450 clocked @ 3.6-3.8 a run for its money. :T

xbit labs showed that quads will have a very hard time passing 450fsb. Dont even think about 500fsb. And 450+ will require a GREAT board to keep and hold for long term.

true but a Q6700 running at 4.0 ghz will suck two times more power and require watercooling.

a Q9450 will run 3.6 ghz with stock voltage and stock cooling, use half the power and perform roughly the same.

not bad eh?

 

[JAG87]

Originally posted by: harpoon84

Originally posted by: aigomorla
Assuming you could keep the temps in check:
Actually im thinkn a GOOD G0 Q6700 clocked @ 4.0ghz would give a Q9450 clocked @ 3.6-3.8 a run for its money. :T

xbit labs showed that quads will have a very hard time passing 450fsb. Dont even think about 500fsb. And 450+ will require a GREAT board to keep and hold for long term.

It certainly would, but how much power would a hypothetical 4GHz Q6700 pull?

Even at 3.6GHz, a Q6600 is already pulling an extra 70W under load compared to a Q9300 @ 3.5GHz. I'm afraid trying to milk that final 400MHz from a 65nm Kentsfield would be akin to you trying to squeeze that final 300MHz from your E8400 - the volts/power would just skyrocket and make it unfeasible to achieve with standard air cooling.

son of a bitch, beat me to it.

 

[harpoon84]

Originally posted by: JAG87
son of a bitch, beat me to it.

No worries bro, you'll pip me another day.

 

[tenax]

done graysky..any feedback you have on those for me by the way is appreciated..at this point, does appear to be an fsb wall on it around 450..tried 7.5 x 475 no dice, but maybe limit of my 6400 ram. we'll find out more monday hopefully and if i get higher, i'll do another run at it and add to the existing results.

 

[tenax]

Originally posted by: JAG87

Originally posted by: harpoon84

Originally posted by: aigomorla
Assuming you could keep the temps in check:
Actually im thinkn a GOOD G0 Q6700 clocked @ 4.0ghz would give a Q9450 clocked @ 3.6-3.8 a run for its money. :T

xbit labs showed that quads will have a very hard time passing 450fsb. Dont even think about 500fsb. And 450+ will require a GREAT board to keep and hold for long term.

It certainly would, but how much power would a hypothetical 4GHz Q6700 pull?

Even at 3.6GHz, a Q6600 is already pulling an extra 70W under load compared to a Q9300 @ 3.5GHz. I'm afraid trying to milk that final 400MHz from a 65nm Kentsfield would be akin to you trying to squeeze that final 300MHz from your E8400 - the volts/power would just skyrocket and make it unfeasible to achieve with standard air cooling.

son of a bitch, beat me to it.

i can do 8 x 450 stable with my q9450 (3.6) at 1.3 volts on the cpu..what would the q6700 need for voltage to do that? I can't read my temps which is driving me nuts due to the bios on my abit not ready for prime time when it comes to reading those unfortunately..i'm hoping abit fixes that..soon

 

[harpoon84]

Originally posted by: tenax
i can do 8 x 450 stable with my q9450 (3.6) at 1.3 volts on the cpu..what would the q6700 need for voltage to do that? I can't read my temps which is driving me nuts due to the bios on my abit not ready for prime time when it comes to reading those unfortunately..i'm hoping abit fixes that..soon

Generally, G0 Q6600s (there ain't many Q6700 users around) need around 1.4V for 3.6GHz. Don't forget that a 3.6GHz Q9450 is equal to a 3.8GHz Q6700 though.

 

[JAG87]

Originally posted by: harpoon84

Originally posted by: tenax
i can do 8 x 450 stable with my q9450 (3.6) at 1.3 volts on the cpu..what would the q6700 need for voltage to do that? I can't read my temps which is driving me nuts due to the bios on my abit not ready for prime time when it comes to reading those unfortunately..i'm hoping abit fixes that..soon

Generally, G0 Q6600s (there ain't many Q6700 users around) need around 1.4V for 3.6GHz. Don't forget that a 3.6GHz Q9450 is equal to a 3.8GHz Q6700 though.

4 ghz

the performance increase of penryn is roughly equal to a 400 mhz gain. you can find a lot of benchmarks to back this, im not making it up. 3.2 ghz penryn = 3.6 ghz kents, and 3.6 penryn = 4.0 kents

 

[harpoon84]

Originally posted by: JAG87
4 ghz

the performance increase of penryn is roughly equal to a 400 mhz gain. you can find a lot of benchmarks to back this, im not making it up. 3.2 ghz penryn = 3.6 ghz kents, and 3.6 penryn = 4.0 kents

From what I've seen, 45nm brings about a 5% IPC increase. At least this is the case between dual cores, and I don't see why that would change with quads.

I don't think a 3.6GHz Penryn would be quite fast enough to match a 4GHz Kentsfield, although I would be happy to be proven wrong. Do you have links to such benchmarks?

 

[JAG87]

Originally posted by: harpoon84

Originally posted by: JAG87
4 ghz

the performance increase of penryn is roughly equal to a 400 mhz gain. you can find a lot of benchmarks to back this, im not making it up. 3.2 ghz penryn = 3.6 ghz kents, and 3.6 penryn = 4.0 kents

From what I've seen, 45nm brings about a 5% IPC increase. At least this is the case between dual cores, and I don't see why that would change with quads.

I don't think a 3.6GHz Penryn would be quite fast enough to match a 4GHz Kentsfield, although I would be happy to be proven wrong. Do you have links to such benchmarks?

you can find some examples on this page

http://www.anandtech.com/cpuch...owdoc.aspx?i=3184&p=12

and it looks like you are right. i must have read that chart wrong before.

 

[Tempered81]

Originally posted by: harpoon84

Originally posted by: JAG87
4 ghz

the performance increase of penryn is roughly equal to a 400 mhz gain. you can find a lot of benchmarks to back this, im not making it up. 3.2 ghz penryn = 3.6 ghz kents, and 3.6 penryn = 4.0 kents

From what I've seen, 45nm brings about a 5% IPC increase. At least this is the case between dual cores, and I don't see why that would change with quads.

I don't think a 3.6GHz Penryn would be quite fast enough to match a 4GHz Kentsfield, although I would be happy to be proven wrong. Do you have links to such benchmarks?

Yeah I agree with the 4-6% IPC increase. It wouldn't make sense to just say 400mhz. that logic would be the same as saying a 100mhz q9450 = a 500mhz q6700.

edit:
at 6% on a given benchmark, that would mean a 3816mhz kentsfield = 3600mhz yorkfield

 

[harpoon84]

Originally posted by: JAG87
you can find some examples on this page

http://www.anandtech.com/cpuch...owdoc.aspx?i=3184&p=12

and it looks like you are right. i must have read that chart wrong before.

Thanks, I've actually seen that article already but I totally forgot about it! It provides a good reference point for the relative performance between overclocked Penryns and Kentsfields.

 

[harpoon84]

Woops double post!

 

[tenax]

cool..would be thrilled if i can get 3.8 with the 9450..we'll see monday i guess