Tom's Hardware > Forum > CPU & Components > CPUs > Performance difference 65nm vs 45nm quad core

Performance difference 65nm vs 45nm quad core

Forum CPU & Components : CPUs - Performance difference 65nm vs 45nm quad core

Tom's Hardware: Over 1.4 million members in 6 different countries available to answer all your high-tech questions. Sign up now! Its free!
Page:    Previous 1 2 Next Bottom Search this thread
Word :    Username :           
 

I've been looking at this system to get and its a really great price but I'm hesitating buying it because of the

new cpu's comming out in the next few months. It comes with a 780i so I could always upgrade but that wouldn't

make much $ sense. Is there a big performance difference between the q6600 and whatever replaces it? Any

thoughts or suggestions would be helpful.



Also I know a few of you are gonna say build my own system but I just don't have the time or skill.

Sponsored Links
Register or log in to remove.

It's something like 7% clock for clock. However the low end lineup on the 45nm is:

Q9300 - 2.5 GHz, 6 MB L2 Cache
Q9450 - 2.66 GHz, 12 MB L2 Cache

Q6600 - 2.4 GHz, 8 MB L2 Cache

So, if you aren't OCing the Q9300 should beat it and be cheaper. The Q9450 will be slightly more expensive and require a better board for OCing (8x versus 9x).

------------------------------ TeamBAG Member
Reply to cnumartyr
- 0 +

if you are happy to OC (and its not hard) the q6600 will still be better, becuase intel are limiting the lower end 45nm CPU's with crappy multipliers

Reply to spuddyt

spuddyt wrote :

if you are happy to OC (and its not hard) the q6600 will still be better, becuase intel are limiting the lower end 45nm CPU's with crappy multipliers



Unless you want to run your board at 500 MHz FSB 24/7!

------------------------------ TeamBAG Member
Reply to cnumartyr

500FSB is not easy with a Quad.
They put far more stress on your Mobo.

------------------------------ If its good in theory but not in practice,
its not good theory.
Reply to zenmaster

zenmaster wrote :

500FSB is not easy with a Quad.
They put far more stress on your Mobo.



Agreed. That was my point.

So.. here is some rampant speculation.

45nm = 7% more Clock for Clock.
Going from 1 MB Cache to 2 MB Cache = 5% more Clock for Clock.

So.. here goes some crazy math that is probably WAY off!

A 100% increase in Cache gets you 5% clock for clock. So moving from 8 MB cache to 12 MB cache would yeild a 50% increase or theoretical 2.5% increase.

Dropping from 8 MB to 6 MB would be a theoretical 1.25% Drop.

So..

2.5 GHz Q9300 = 2.64 GHz Q6600 (at 333 MHz FSB the Q6600 would be at 3.0 GHz)
2.66 GHz Q9450 = 2.98 GHz Q6600 (at 333 MHz FSB the Q6600 would be at 3.0 GHz)

So.. on to FSB multipliers.

400 MHz FSB!

3000 MHz Q9300 = 3.17 GHz Q6600 (3.6 GHz at 400 MHz FSB)
3200 MHz Q9450 = 3.5 GHz Q6600 (3.6 GHz at 400 MHz FSB)


The Q6600 will be hotter and require more cooling and more voltage at all levels. The question is how high do you have to run the Q9450 to beat the Q6600.

------------------------------ TeamBAG Member
Reply to cnumartyr
- 0 +

According to the non-synthetic THG benchmarks (22 tests) of the qx9650 (12mb cache) vs. qx6850 (8mb cache) which are both clocked at 3GHz using 333 FSB...

...the 45nm averages 5.2% better than 65nm, clock for clock, not 7%.

so... (assuming q6600 and qx6850 clocks are comparable)

2.5 GHz Q9300 = 2.63 GHz Q6600 (9x262)
2.66 GHz Q9450 = 2.81 GHz Q6600 (9x312)

Quote :

So.. on to FSB multipliers.

400 MHz FSB!



3000 MHz Q9300 = 3.16 GHz Q6600 (9x351)
3200 MHz Q9450 = 3.37 GHz Q6600 (9x374)

Reply to coret

Then I was wrong and it's an even better reason to keep using G0 Q6600s.

------------------------------ TeamBAG Member
Reply to cnumartyr
- 0 +

Ignore above post ... I did the maths wrong >.<

I shouldn't try to do maths when i've had beers...

Reply to coret
- 0 +

coret wrote :

Ignore above post ... I did the maths wrong >.<

I shouldn't try to do maths when i've had beers...



And as a sign that I'm not very good doing maths when I've had beer, my housemate just pointed out I got it right.... Ignore that post :p

Reply to coret

coret wrote :

And as a sign that I'm not very good doing maths when I've had beer, my housemate just pointed out I got it right.... Ignore that post :p




You did it right, but you forgot ONE thing. On the Q9300. It's 6 MB of L2 Cache and not the 12 MB of L2 Cache that the QX9650 had in the test... so I would say it's clock for clock advantage would be even lower.

------------------------------ TeamBAG Member
Reply to cnumartyr

So in conclusion, just get the Q6600 and overclock it to heaven.

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri
- 0 +

cnumartyr wrote :

You did it right, but you forgot ONE thing. On the Q9300. It's 6 MB of L2 Cache and not the 12 MB of L2 Cache that the QX9650 had in the test... so I would say it's clock for clock advantage would be even lower.



Good point ... long live the G0 Q6600! (even though I don't have one...)

I did get one thing wrong though ... a typo on "2.5 GHz Q9300 = 2.63 GHz Q6600 (9x262)" should be 9x292.

Reply to coret

Evilonigiri wrote :

So in conclusion, just get the Q6600 and overclock it to heaven.




Yea..

For 24/7 OCs I would say the Q6600 would be better since you can hit 3.6 GHz on Air without a problem. You'd need 450 MHz FSB on that for the Q9450... At the same time you'd be pushing the board pretty hard. If you were doing water you'd have the Q6600 at 4 GHz.

It's all a trade off.

------------------------------ TeamBAG Member
Reply to cnumartyr

coret wrote :

According to the non-synthetic THG benchmarks (22 tests) of the qx9650 (12mb cache) vs. qx6850 (8mb cache) which are both clocked at 3GHz using 333 FSB...

...the 45nm averages 5.2% better than 65nm, clock for clock, not 7%.




You want to split hairs over 1.8%?! :lol:

OK, since we're nitpicking, how about the fact that the THG tests had both 45nm and 65nm at the same FSB, which wouldn't be the case when comparing a Q9300/Q9450 against the Q6600.

If you are talking a Q9450 vs Q6600 @ 3GHz, it would be a 375MHz FSB vs 333MHz, a 12.5% difference in FSB speed. Which probably accounts for about 1% of your 1.8% deficit, now do we want to argue over that final 0.8%? :kaola:

Reply to epsilon84

The Q6600 is an excellent choice. As previously stated they can be greatly overclocked. They also will probably do whatever you need one to do even at stock speeds. The new CPUs I believe will provide even better preformance. To what degree we shall see and they might even have a better price v preformance ratio. If you jump now do so knowing you'll have a fine CPU and if you do don't sweat it if the upcoming CPUs are a bit better. We can't chase the tech dragon because we'll never catch it. A middle road would be to wait (and we hate to wait) to see how the new chips review and preform; compare price and make your most reasoned decision. As a side note: by waiting it is likely that some of the rest of your parts might be a bit cheaper.

Reply to topper743

cnumartyr wrote :

Yea..

For 24/7 OCs I would say the Q6600 would be better since you can hit 3.6 GHz on Air without a problem. You'd need 450 MHz FSB on that for the Q9450... At the same time you'd be pushing the board pretty hard. If you were doing water you'd have the Q6600 at 4 GHz.

It's all a trade off.



Ahh, I love debating this with you cnumartyr! :p

You are right, even at 450FSB you'd be pushing the mobo pretty hard. But conversely, you'd hardly be pushing the CPU. In fact you might even do it at stock volts!

To get a Q6600 to equivalent levels (~3.8GHz) you would generally need to push the CPU *very* hard.

I've sure I've said this before but I'll say it again - I don't think the Q9450 is the best choice for hardcore overclockers. 4GHz is definitely unlikely. Even with the Q9550, 4GHz wouldn't be guaranteed, since you'll still need a 470MHz FSB. But for anybody who wants equivalent to slightly better performance than a Q6600 @ 3.6GHz whilst halving the CPU power consumption, the Q9450 is a good choice.

Reply to epsilon84

epsilon84 wrote :

Ahh, I love debating this with you cnumartyr! :p

You are right, even at 450FSB you'd be pushing the mobo pretty hard. But conversely, you'd hardly be pushing the CPU. In fact you might even do it at stock volts!

To get a Q6600 to equivalent levels (~3.8GHz) you would generally need to push the CPU *very* hard.

I've sure I've said this before but I'll say it again - I don't think the Q9450 is the best choice for hardcore overclockers. 4GHz is definitely unlikely. Even with the Q9550, 4GHz wouldn't be guaranteed, since you'll still need a 470MHz FSB. But for anybody who wants equivalent to slightly better performance than a Q6600 @ 3.6GHz whilst halving the CPU power consumption, the Q9450 is a good choice.




What are we debating? We agree! :kaola:

For WC setups and stuff like that I think the Q6600 is a better choice to push it further.

The Q9450 is great for non-extreme OCers with 3.2 GHz. I think the sweet spot will be 3.2 GHz with 1600 MT/s FSB.

But again.. At the same FSB the Q6600 will always win. The Q9450 will be a great chip. Even the Q9300 will.

I am planning on getting another 2-3 Q6600s to find a 1.212 VID or so.

------------------------------ TeamBAG Member
Reply to cnumartyr

cnumartyr wrote :

For WC setups and stuff like that I think the Q6600 is a better choice to push it further.


Ah...the water closet?

Sorry, I always see it as that. :(

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri

Evilonigiri wrote :

Ah...the water closet?

Sorry, I always see it as that. :(



Are you from Germany?

------------------------------ TeamBAG Member
Reply to cnumartyr

cnumartyr wrote :

Are you from Germany?


No I'm from Hawaii :pt1cable:

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri

Evilonigiri wrote :

No I'm from Hawaii :pt1cable:




Lol.. I knew they called it a Water Closet in Germany, don't know much about Hawaii, I'm on the East Coast. :kaola:

------------------------------ TeamBAG Member
Reply to cnumartyr

cnumartyr wrote :

Lol.. I knew they called it a Water Closet in Germany, don't know much about Hawaii, I'm on the East Coast. :kaola:


Currently, I'm on the West Coast. I lived in Hawaii for 10years.

Must be hanging around too many German people...

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri
- 0 +

epsilon84 wrote :

You want to split hairs over 1.8%?! :lol:

When dealing with CPU's doing billions of calculations per second, and how performance/power need/heat production varies with OCing, yes, I will split hairs over 1.8% :)

Quote :

OK, since we're nitpicking, how about the fact that the THG tests had both 45nm and 65nm at the same FSB, which wouldn't be the case when comparing a Q9300/Q9450 against the Q6600.



Which would be precisely the reason I chose equally clocked chips with the same FSB - otherwise it's not a fair test.

Also, if you check, I said this... "(assuming q6600 and qx6850 clocks are comparable)" to help cover for that exact problem.

Quote :

If you are talking a Q9450 vs Q6600 @ 3GHz, it would be a 375MHz FSB vs 333MHz, a 12.5% difference in FSB speed. Which probably accounts for about 1% of your 1.8% deficit, now do we want to argue over that final 0.8%? :kaola:



Find me a benchmark showing that 1% you claim and I'll agree with it :)

Can I go to bed now without having my reasonably valid calculations + assumptions mocked in my sleep? :sleep:

Reply to coret

Evilonigiri wrote :

Currently, I'm on the West Coast. I lived in Hawaii for 10years.

Must be hanging around too many German people...



I think so.. I spent a couple weeks in Germany and that's all I heard was the WC...

East Coast > West Coast. :kaola:

------------------------------ TeamBAG Member
Reply to cnumartyr

Not to end the big debate here or anything but I'm not that big on overclocking. The most overclocking I've done is

in nTune and it didn't support my cpu at the time.

Message quoted 1 times
Message edited by gamecrazychris on 01-20-2008 at 03:01:42 AM
Reply to gamecrazychris

gamecrazychris wrote :

Not to end the big debate here or anything but I'm not that big on overclocking. The most overclocking I've done is in nTune and it didn't support my cpu at the time.




At completely stock, than yes the Q6600 is beat by it's replacement Q9450. A Q9450 at stock will perform roughly like a 3.0 GHz Q6600 (which coincidentally is a 333x9 Q6600).

Edit: But there are a lot of us overclockers here. :kaola:

Message quoted 1 times
Message edited by cnumartyr on 01-20-2008 at 03:01:00 AM
------------------------------ TeamBAG Member
Reply to cnumartyr

cnumartyr wrote :

I think so.. I spent a couple weeks in Germany and that's all I heard was the WC...

East Coast > West Coast. :kaola:


What, in OCing? :kaola:

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri

cnumartyr wrote :

Edit: But there are a lot of us overclockers here. :kaola:


The OCing population is 2% I believe. That's not a lot...

The OCing population here on TGH is like 80% if that's what you meant.

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri

Evilonigiri wrote :

The OCing population is 2% I believe. That's not a lot...

The OCing population here on TGH is like 80% if that's what you meant.



That's exactly what I meant. I would venture to say that most people on Toms have SOME experience with OCing.

------------------------------ TeamBAG Member
Reply to cnumartyr

Its not that I'm against overclocking, I just don't want to risk breaking

my new system. If anyone knows of a good overclocking video please

tell me. If I did decide to overclock it then it wouldn't be that extreme

mabey to 3.1 ghz

Reply to gamecrazychris

gamecrazychris wrote :

Its not that I'm against overclocking, I just don't want to risk breaking

my new system. If anyone knows of a good overclocking video please

tell me. If I did decide to overclock it then it wouldn't be that extreme

mabey to 3.1 ghz



If you were OCing a Q6600 from 266x9 to 333x9 it would be on par with a stock Q9450 and be cheaper. You could also get it now rather than waiting.

For 3 GHz on a Q9450 needs a 375 MHz FSB...

Go to the OC section of this website and read up on C2D and C2Q OCing. If you have any questions we have a great community here that will be willing to help you.

Edit: My Q6600 sits at 3.6 GHz daily. I've had it up to 3.89 for Benches on air.

Message quoted 1 times
Message edited by cnumartyr on 01-20-2008 at 03:07:30 AM
------------------------------ TeamBAG Member
Reply to cnumartyr

gamecrazychris wrote :

Its not that I'm against overclocking, I just don't want to risk breaking

my new system. If anyone knows of a good overclocking video please

tell me. If I did decide to overclock it then it wouldn't be that extreme

mabey to 3.1 ghz


There is a great overclocking guide here. You might want to check that out.

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri
- 0 +

cnumartyr wrote :

I've had it up to 3.89 for Benches on air.



What time do you get in SuperPI 1million at that speed?

Reply to coret

In real life situations will I really notice a difference between 2.6ghz and

3ghz? Also overclocking voids the warrenty on my new computer so if

something does happen I would pretty much be screwed right?

Reply to gamecrazychris

coret wrote :

What time do you get in SuperPI 1million at that speed?



I didn't SuperPi, I was 3DMark06ing.

My best SuperPi1M was 14.09 with DDR2-1200 and 3.6 GHz. And no I did no Windows Tweaks or anything. It was just a straight up install like normal.

------------------------------ TeamBAG Member
Reply to cnumartyr

gamecrazychris wrote :

In real life situations will I really notice a difference between 2.6ghz and

3ghz? Also overclocking voids the warrenty on my new computer so if

something does happen I would pretty much be screwed right?




Well, if you buy like a Dell or something you won't be able to OC it. Most of us build our own PCs and OC them. OEMs (Dell, HP) normally lock their BIOS's so you can't OC it.

Yes it voids the warranty. However I've gotten motherboards, RAM, CPUs, and Video cards that I've OC'd replaced. If I even see it doing something I don't like I return it. I RMAed a motherboard because I needed 1.4v for 333 MHz on the Northbridge. They took it back and replaced it.

Anyways.. if you are buying it from a Dell or something, I would recommend waiting for a Q9450.

Edit: Also what is a real life situation for you? Gaming? Web browsing? Rendering / Editing or Encoding Video or MP3s?


Message edited by cnumartyr on 01-20-2008 at 03:13:59 AM
------------------------------ TeamBAG Member
Reply to cnumartyr

gamecrazychris wrote :

In real life situations will I really notice a difference between 2.6ghz and

3ghz? Also overclocking voids the warrenty on my new computer so if

something does happen I would pretty much be screwed right?


They won't know, unless you told them. ^^

And yes you will see a difference in 2.6Ghz and 3.0Ghz. I did, at least in supcom.

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri
- 0 +

gamecrazychris wrote :

In real life situations will I really notice a difference between 2.6ghz and

3ghz? Also overclocking voids the warrenty on my new computer so if

something does happen I would pretty much be screwed right?



You probably won't notice a difference unless you do a vast quantity of number crunching like video encoding or graphics rendering or something along those line.

Overclocking may void the warranty, but I've yet to see a way to tell whether or not chip has been overclocked other than being scarred by heat (which doesn't happen anymore as CPU's throttle back to avoid heat damage)

Plus when overclocking, a couple of boots at unstable clock speeds won't hurt it ... just reset the CMOS (if it doesn't automatically) and you're back to where you were before :)

Reply to coret

coret wrote :

You probably won't notice a difference unless you do a vast quantity of number crunching like video encoding or graphics rendering or something along those line.

Overclocking may void the warranty, but I've yet to see a way to tell whether or not chip has been overclocked other than being scarred by heat (which doesn't happen anymore as CPU's throttle back to avoid heat damage)

Plus when overclocking, a couple of boots at unstable clock speeds won't hurt it ... just reset the CMOS (if it doesn't automatically) and you're back to where you were before :)




Let's see what I've done to my stuff that STILL works.

I've frozen my computer atleast 30-40 times while OCing my GPU. Trying to find the max I can possibly finish 3DMark06 at. Nothing happened, everything is still good.

I've crashed I can't tell you how many times from memory problems during OCing it.

I've crashed my computer a tons of times OCing my CPU and getting auto reboots.

I ran my P4 @ 3.2 GHz with no thermal grease for 4 months and it shut down 3-4 times a day. I was just too lazy to open it up and put more on. It's still chuggin strong.

Edit: I've run my Q6600 at over 1.6v on Air. It's still fine.

Message quoted 2 times
Message edited by cnumartyr on 01-20-2008 at 03:17:21 AM
------------------------------ TeamBAG Member
Reply to cnumartyr

If you want to know I was looking at

http://www.cyberpowerpc.com/system [...] urator_top

I know that they don't have the best track reccord but I changed

some stuff and the price turned out pretty good

Reply to gamecrazychris

cnumartyr wrote :

Let's see what I've done to my stuff that STILL works.

I've frozen my computer atleast 30-40 times while OCing my GPU. Trying to find the max I can possibly finish 3DMark06 at. Nothing happened, everything is still good.

I've crashed I can't tell you how many times from memory problems during OCing it.

I've crashed my computer a tons of times OCing my CPU and getting auto reboots.

I ran my P4 @ 3.2 GHz with no thermal grease for 4 months and it shut down 3-4 times a day. I was just too lazy to open it up and put more on. It's still chuggin strong.

Edit: I've run my Q6600 at over 1.6v on Air. It's still fine.


I've got 2 words for you...

STOP BRAGGING.

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri

gamecrazychris wrote :

If you want to know I was looking at

http://www.cyberpowerpc.com/system [...] urator_top

I know that they don't have the best track reccord but I changed

some stuff and the price turned out pretty good



If you were comfortable with it (and most of us would help you with component selection and compatibility) you could build your own for a lot cheaper with better parts.

------------------------------ TeamBAG Member
Reply to cnumartyr

Evilonigiri wrote :

I've got 2 words for you...

STOP BRAGGING.



Lol, how is it bragging talking about the stuff I've done to my stuff and crash my computer multiple times?

------------------------------ TeamBAG Member
Reply to cnumartyr

cnumartyr wrote :

Lol, how is it bragging talking about the stuff I've done to my stuff and crash my computer multiple times?


"I crashed a lot but it's still fine" would have been just fine.

No need the details.

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri

I'll build my next computer but I just can't at the moment. I just moved

and It'll take me at least 3 months until I have enough free time to build

my own system. But back to the topic, I guess I'll just wait like 2

months.

Reply to gamecrazychris

Evilonigiri wrote :

"I crashed a lot but it's still fine" would have been just fine.

No need the details.



Ok, I'll give you that much. However I was just trying to give details so the OP knew just how many times I have messed up and not damaged any equipment.

------------------------------ TeamBAG Member
Reply to cnumartyr

gamecrazychris wrote :

I'll build my next computer but I just can't at the moment. I just moved

and It'll take me at least 3 months until I have enough free time to build

my own system. But back to the topic, I guess I'll just wait like 2

months.



That's a good idea, and by then you can build it yourself!

And sorry for hijacking your thread with Evil, we do that alot. He and I need to just exchange Steam IDs or AIM.

------------------------------ TeamBAG Member
Reply to cnumartyr

cnumartyr wrote :

Ok, I'll give you that much. However I was just trying to give details so the OP knew just how many times I have messed up and not damaged any equipment.


With an ulterior motive... :whistle:

------------------------------ "Nvidia, the Way It's Meant to be PAID Played! - Corrado
*Lesbian Lover Club* - founder Assman
Reply to Evilonigiri

Oh crap I forgot, I was gonna get an nvidia 9800 gx2 when it comes out

in Febuary-March so I guess I have no option but to wait.

Anyway thanks for all the info. I'll come back to the fourms when I get

my new system and if I decide to overclock it.

Reply to gamecrazychris
Previous
1 2
Tom's Hardware > Forum > CPU & Components > CPUs > Performance difference 65nm vs 45nm quad core
Go to:

There are 1111 identified and unidentified users. To see the list of identified users, Click here.

Please mind

You are about to answer a thread that has been inactive for more than 6 months.
If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.

Add a reply Cancel
Sponsored links
  • Ask the community now
  • Publish
Ad
They won a badge
Join us in greeting them
  • 01:00 bew1075k won the Uniformed badge
  • 01:00 Parasitic won the Uniformed badge
  • 01:00 X3no won the Uniformed badge
  • 01:00 Siq won the Uniformed badge
  • 01:00 wayspooled won the Uniformed badge
  • 01:00 chris62 won the Freshman badge
  • 01:00 Tattysnuc won the Freshman badge
  • 01:00 gamerk316 won the Freshman badge
  • 05:23 jsc won the Overclocking badge
  • 01:00 rolli59 won the Spy badge