Q9300 reviewed

[harpoon84]

http://www.xbitlabs.com/articl...uad-q9300_2.html#sect0

Compared to the Q6600, it's about 7% faster. Unfortunately, the 7.5x multiplier limits its overclocking potential, and 3.5GHz is about as far as you can go before hitting a FSB wall.

The low multiplier aside, it's not a bad CPU, considering it's the same price as a Q6600. It doesn't blow the Q6600 away, but it is slightly faster and lower power consumption is always welcome.

 

[NinjaJedi]

I wonder if the Q9450 is going to have a higher multiplier. The Q9300 only has the 6mb cache. I'm thinking the Q9300 has 7 Q9450 has 8 and Q9550 will have the 9. Seems like a decent bump up in price for double the cache and a higher multi if the Q9450 has 8.

 

[harpoon84]

Originally posted by: NinjaJedi
I wonder if the Q9450 is going to have a higher multiplier. The Q9300 only has the 6mb cache. I'm thinking the Q9300 has 7 Q9450 has 8 and Q9550 will have the 9. Seems like a decent bump up in price for double the cache and a higher multi if the Q9450 has 8.

Actually, the multis increase in .5 increments

Q9300 - 7.5x / 2.5GHz
Q9450 - 8x / 2.66GHz
Q9550 - 8.5x / 2.83GHz

The price difference between the Q9300 and Q9450 ain't much, about $50, so yeah, I'd probably choose the Q9450. Twice the cache and slightly higher overclocking headroom.

However, I definitely wouldn't spend an extra $200 just for the .5x higher multi on the Q9550.

 

[exar333]

If I remember right, the Q9450 does have a 8 multipler, but the Q9550 has a 8.5, not a 9...

Edit: LOL, beat me to it!

 

[NinjaJedi]

I was under the impression that to get a multi of .5 you also need one of the newer M/B (x38/x48)? That or I just got confused like always.

 

[harpoon84]

Originally posted by: NinjaJedi
I was under the impression that to get a multi of .5 you also need one of the newer M/B (x38/x48)? That or I just got confused like always.

I don't think so, in fact even older P965 boards can run 45nm chips with an updated BIOS. P35 certainly won't have a problem since it's designed for 45nm anyway, you definitely don't need an X38/X48 mobo.

 

[JAG87]

yippeee, just like I predicted. no high clocks for mainstream chips this time around.

 

[harpoon84]

Originally posted by: JAG87
yippeee, just like I predicted. no high clocks for mainstream chips this time around.

Well, if the Q9450 has a similar FSB limit to the Q9300, then it'll top out around 3.7 - 3.8GHz, which ain't far from your QX9650 @ 4GHz, for 1/3 the price I'd probably take the Q9450.

 

[imported_Scoop]

Well that's pretty nice, it overcomes the cache loss without a sweat. The lower multiplier is only an issue if you get a bad board, not the chips fault.

 

[harpoon84]

Originally posted by: Scoop
Well that's pretty nice, it overcomes the cache loss without a sweat. The lower multiplier is only an issue if you get a bad board, not the chips fault.

Most P35/X38 boards top out around ~475FSB though, so basically you won't get above 3.6GHz with a Q9300 unless you have an EXCEPTIONAL mobo.

 

[NinjaJedi]

It would be nice if a 400 FSB would not need the volt increase like the e8400 do.

 

[nyker96]

nice, from the looks of it, my next upgrade might be E8xxx since quad support isn't as prevalent yet from the benches. Also I'd stay away from Q6600s now that I can see the power consumption is so much more at load/idle. Assumming 40W more/hour of usage, I'm bleeding $25 a year into electricity bills, definitely going 45nm next.

 

[VirtualLarry]

Even at 3.5Ghz, it beats out the Q6600 on many apps at 3.6Ghz.

 

[JAG87]

Originally posted by: harpoon84

Originally posted by: JAG87
yippeee, just like I predicted. no high clocks for mainstream chips this time around.

Well, if the Q9450 has a similar FSB limit to the Q9300, then it'll top out around 3.7 - 3.8GHz, which ain't far from your QX9650 @ 4GHz, for 1/3 the price I'd probably take the Q9450.

thats true, but you wont have the best for cheap this time like it was common with the q6600. plus it sucks balls to know that your cpu could easily do 4.0, 4.1 or even 4.2 ghz but you are stuck at 3.8 ghz, if you even make it there. Xbit labs used a P5E3 Deluxe motherboard + DDR3 memory, not exactly cheap and not exactly what budget conscious Q9450 owners will consider "worth the money".

my post was merely made to point out the fsb limitation. there have been a lot of posts lately about wofldales and yorkfields, and how high they will be able to clock. I think JaredPace himself must have created 3 threads on this subject. Well I hope this Xbit article cleared things up for most people.

 

[aigomorla]

Originally posted by: harpoon84

Originally posted by: JAG87
yippeee, just like I predicted. no high clocks for mainstream chips this time around.

Well, if the Q9450 has a similar FSB limit to the Q9300, then it'll top out around 3.7 - 3.8GHz, which ain't far from your QX9650 @ 4GHz, for 1/3 the price I'd probably take the Q9450.

yeah but you'll end up replacing that board pretty quickly... roughly around 6-8months for an RMA date, from all that stress.

The problem isnt getting the FSB, its holding it from the stress. That requirement would force you to add some major voltages on the NB, FSB Termination.

:T


So if i had to factor in the downtime on the motherboard swap, the QX9650 is a more feasible soltion if your going at upper area overclocking.

 

[AdamK47]

Wow! I didn't realize just how much of a power hog a Q6600 @ 3.6GHz was. My E8500 @ 4.05GHz would probably be around 275W compared to the 393.7W of the Q6600 @ 3.6GHz.

 

[harpoon84]

Originally posted by: aigomorla

Originally posted by: harpoon84

Originally posted by: JAG87
yippeee, just like I predicted. no high clocks for mainstream chips this time around.

Well, if the Q9450 has a similar FSB limit to the Q9300, then it'll top out around 3.7 - 3.8GHz, which ain't far from your QX9650 @ 4GHz, for 1/3 the price I'd probably take the Q9450.

yeah but you'll end up replacing that board pretty quickly... roughly around 6-8months for an RMA date, from all that stress.

The problem isnt getting the FSB, its holding it from the stress. That requirement would force you to add some major voltages on the NB, FSB Termination.

:T


So if i had to factor in the downtime on the motherboard swap, the QX9650 is a more feasible soltion if your going at upper area overclocking.

Are you speaking from personal experience? Have you had boards fail due to prolonged periods at a high FSB?

 

[aigomorla]

Originally posted by: harpoon84

Originally posted by: aigomorla

Originally posted by: harpoon84

Originally posted by: JAG87
yippeee, just like I predicted. no high clocks for mainstream chips this time around.

Well, if the Q9450 has a similar FSB limit to the Q9300, then it'll top out around 3.7 - 3.8GHz, which ain't far from your QX9650 @ 4GHz, for 1/3 the price I'd probably take the Q9450.

yeah but you'll end up replacing that board pretty quickly... roughly around 6-8months for an RMA date, from all that stress.

The problem isnt getting the FSB, its holding it from the stress. That requirement would force you to add some major voltages on the NB, FSB Termination.

:T


So if i had to factor in the downtime on the motherboard swap, the QX9650 is a more feasible soltion if your going at upper area overclocking.

Are you speaking from personal experience? Have you had boards fail due to prolonged periods at a high FSB?

2 680i's and a P35 is starting to show wear as we speak.

 

[n7]

Gotta love Intel for this.

We all knew we this was coming.

Now if we want serious clocking, we must shell out for the uber priced quads.

Not that this is news really...just now it's confirmation of what we knew to be coming.

 

[Markfw]

Originally posted by: harpoon84

Originally posted by: aigomorla

Originally posted by: harpoon84

Originally posted by: JAG87
yippeee, just like I predicted. no high clocks for mainstream chips this time around.

Well, if the Q9450 has a similar FSB limit to the Q9300, then it'll top out around 3.7 - 3.8GHz, which ain't far from your QX9650 @ 4GHz, for 1/3 the price I'd probably take the Q9450.

yeah but you'll end up replacing that board pretty quickly... roughly around 6-8months for an RMA date, from all that stress.

The problem isnt getting the FSB, its holding it from the stress. That requirement would force you to add some major voltages on the NB, FSB Termination.

:T


So if i had to factor in the downtime on the motherboard swap, the QX9650 is a more feasible soltion if your going at upper area overclocking.

Are you speaking from personal experience? Have you had boards fail due to prolonged periods at a high FSB?

What is considered a high FSB ? I have a plain S3 board thats been doing 400 on a Q6600 for 9 months now, 24/7, no problem. The highest I have is 415, but thats on a DQ6, and I doubt it will ever die, even if it was running 475.

 

[lopri]

High FSB does stress board more (you need more voltages for NB, a simple fact) and unfortunately Intel CPUs don't have on-die NB. Mobo chipsets aren't as resilient as CPUs, and with all the circuits around the chip get more stressed as well. This is why I despise Intel's strategy, despite their excellent CPUs and product launches. From what I have tested, Intel's core logic hit its limit with 975X. (that's over 2 years ago now) The best performance with C2D/Q and DDR2 is, as AT demonstrated in recent X48 review, obtained around 400FSB under 266 strap (or tRD=5). And 4:5 memory configuration if possible. (This is an opinion based on my personal experiences, but I believe it's in line with AT's article)

And if you think about it, we did that already with 975X. Not with quads, but with duals. In other words, we don't really need a 333/400 FSB CPUs. Intel could have just enhanced 975X for quads and fixed memory dividers. Obviously that doesn't do much good for Intel's chipset division. Solution: Make chipsets with better manufacturing so they can clock better. Loosen the timings if it gets in the way. Make sure old chipsets don't function with new CPUs. If that means introducing half multies, then be it!

450FSB is too high even for duals, IMO. Other than some cherry-picked boards, that'll require a looser chipset timings and at the same time higher voltage. I did 500 FSB on P5E-VM HDMI for fun last week, and I think the board is half-dead already. It needs a 24-hour rest before it decides to boot again. :laugh:

 

[Idontcare]

Lopri your arguments against Intel's practices need to be considered with the fact that you are arguing for Intel's to have a strategy solely to accomodate folks who rely on high FSB for overclocking (i.e. people who don't want to pay for what they get).

This is in contrast to Intel's viable strategy of ensuring a lot of secondary companies get to rely on an upgrade cycle of refreshes by people who are willing to spend money.

No surprise at all why Intel does what it does. It doesn't help one tiny niche of their world, us free-loading overclockers, but it does help put the food on the tables of thousands of mobo manufacturer employees.

Of course we don't need 333 or 400MHz FSB's. 266 was pretty much the endpoint of point of diminishing returns. But higher and higher FSB gives the memory manufacturers a market to sell the next speedbin of DDR2 as we migrated from 533->667->800->1066.

And its not about screwing us either, the Q9300 is not on 333MHz FSB to screw the enthusiast, it is there so HP and DELL can sell this years quad-core as being billed as superior to last years Q6600 on 267MHz FSB.

It's all about the upgrade cycle, very little if any is about the overclocking enthusiast who wants 2X the performance while spending half as much.

 

[Cheex]

Originally posted by: aigomorla

2 680i's and a P35 is starting to show wear as we speak.

How can you tell?
I'd like to know this bit of information.

 

[lopri]

Originally posted by: Idontcare
Lopri your arguments against Intel's practices need to be considered with the fact that you are arguing for Intel's to have a strategy solely to accomodate folks who rely on high FSB for overclocking (i.e. people who don't want to pay for what they get).

This is in contrast to Intel's viable strategy of ensuring a lot of secondary companies get to rely on an upgrade cycle of refreshes by people who are willing to spend money.

No surprise at all why Intel does what it does. It doesn't help one tiny niche of their world, us free-loading overclockers, but it does help put the food on the tables of thousands of mobo manufacturer employees.

Of course we don't need 333 or 400MHz FSB's. 266 was pretty much the endpoint of point of diminishing returns. But higher and higher FSB gives the memory manufacturers a market to sell the next speedbin of DDR2 as we migrated from 533->667->800->1066.

And its not about screwing us either, the Q9300 is not on 333MHz FSB to screw the enthusiast, it is there so HP and DELL can sell this years quad-core as being billed as superior to last years Q6600 on 267MHz FSB.

It's all about the upgrade cycle, very little if any is about the overclocking enthusiast who wants 2X the performance while spending half as much.

I guess that's that.. Still bitter, though.

 

[marrr]

i was thinking about upgrading so should i get a q6600 or wait for the 45nm?? im looking to overclock.