Quick comparison Q6600/HD7850 Vs i7-3770

Ultra_Extreme · 7 Aug 2012 at 23:39

Ok so since im upgrading thought id post some info, and since my 7850 is faulty ill post more results using a 7950 tomorrow

The purpose hopefully is to assist others in the decision to upgrade to i7 from earlier non 'i' cpus, glad i did!

Q6600 3.0Ghz 4gig hyperX ddr2 @ 800mhz Asus P5QL-Pro

3dmark01 stock 30630
3dmark 03 stock 69680
3dmark06 stock 12502
3dmark vantage stock P13168 16946/7890
3dmark2011 stock P4282 5079/2953/2855
Heaven Stock 1919

Upgrade (i7 3770k auto OC seems to like 4.2ghz under load - 8gig gskill ripjaws 1600mhz Asus P8Z88-V Pro)

3dmark01 stock 72309
3dmark 03 stock 108573
3dmark06 stock 29132
3dmark vantage stock 23530 22040/29515
3dmark2011 stock P6008 5596/9915/5784
Heaven Stock 1772 42 134.9 70.3 av

Interesting results, the vast increase in CPU power shows in all tests up to 3DM2011 where it balances but CPU turns out 9k while gpu can only cope with 5.7k will be interested to see what changes with the 7950 tomorrow. Dont know why Heaven is lower it crashed 4/5 times (because of the faulty GPU) so will write it off for now,

All in all couldnt recommend the upgrade more! Also my SSD running now at 6Gb/s is insane, general performance through the roof, thought an i73770k was overkill versus a 2500k i5 but very glad i splashed out.

Ultra_Extreme · 8 Aug 2012 at 17:56

And the 7950 results

7950 @ stock i7 @ 4.4

3dmark01 74154
3dmark 03 108659
3dmark06 OC 31735
3dmark vantage 28754 28161/30696
3dmark2011 6959 6620/10013/6484
Heaven 2145

Ultra_Extreme · 8 Aug 2012 at 17:58

On the whole id say theres a lot to be said for balancing CPU performance with GPU, but not that much value in the small jump from 7850 - 7950 although i will see better in higher resoloutions which i havent benched. Also GPU still at stock....

Gregster · 8 Aug 2012 at 18:12

Nice info and I am sure will be usefull.

greenpanda · 8 Aug 2012 at 21:43

damn... this is making me want to upgrade

Ultra_Extreme · 8 Aug 2012 at 22:01

huge diff m8 no regrets, thought i would cos i was going a more modest route until i got drunk

figured id get latent buyer remorse but nope this rig will do me a while.

Ultra_Extreme · 8 Aug 2012 at 22:19

Got a Sabretooth Z77 comin tomoz to replace p8z77-v pro so once its in gonna go on overclock mission and get one final set of results.

MeMeMeMe · 9 Aug 2012 at 04:47

I'd have preferred seeing actual gaming benchmarks.

BaDBoY_uK · 9 Aug 2012 at 07:59

A newer CPU gets a better score in benchmarking??? Shocking

Ultra_Extreme · 9 Aug 2012 at 09:33

BaDBoY_uK said:
A newer CPU gets a better score in benchmarking??? Shocking

There are a lot of people out there on older quads probably wondering about the merits of jumping from say 3.2ghz clocked Q series CPU's to 3.4ghz clocked i7 cpus, on paper it doesnt seem worth the hassle. I just wanted to show the results. But then you're a hater badboy_uk lol whats the betting that you're not?

Gregster · 9 Aug 2012 at 10:50

BaDBoY_uK said:
A newer CPU gets a better score in benchmarking??? Shocking

You fail to miss the OP's point!

It is a good reference to somebody who may just want to upgrade a GPU for the time being and at least they can see what kind of bottleneck will happen.

Ultra_Extreme · 9 Aug 2012 at 11:01

gregster said:
You fail to miss the OP's point!

It is a good reference to somebody who may just want to upgrade a GPU for the time being and at least they can see what kind of bottleneck will happen.

thanks for being less ragey than me, he really is missing the point, of course i knew adding £700+ of gear to my system would speed it up, but the point was i was only going to upgrade my 4850 to a 7850 which on reflection now would have been a huge waste of a 7850

Ultra_Extreme · 9 Aug 2012 at 11:03

interestingly note in the run of 3dmark 11 in the last set with the 7950, the cpu gains almost 1.1k in score when only change is gfx card!

Gregster · 9 Aug 2012 at 11:14

Ultra_Extreme said:
interestingly note in the run of 3dmark 11 in the last set with the 7950, the cpu gains almost 1.1k in score when only change is gfx card!

I would have liked to see the 7950 coupled with the Q6600 and a run of 3Dmark11. Often I see people ask "will I get a bottleneck with Qxxxx chip and a 79xx or 6xx". It would have been nice to give them a rough % of lost GPU power but all good.

The hard part is certain games rely more on CPU and others rely more on GPU.

psychodil · 9 Aug 2012 at 11:22

3dmark isn't an accurate depiction of performance lost among typical games though. An over clocked sandy bridge will Max out any single graphics card in games but will not give a high score like the ivy bridges and the hyper threaded processors. In fact I would say that for most games (except a few RTS games) you couldn't notice the difference between a stock 2500k and an over clocked 3970k.

Gregster · 9 Aug 2012 at 11:44

psychodil said:
3dmark isn't an accurate depiction of performance lost among typical games though. An over clocked sandy bridge will Max out any single graphics card in games but will not give a high score like the ivy bridges and the hyper threaded processors. In fact I would say that for most games (except a few RTS games) you couldn't notice the difference between a stock 2500k and an over clocked 3970k.

I don't disagree at all and infact I was scratching my head thinking 'how on earth are these 670's beating me' in the 3dmark11 thread. It was only after looking at the gfx scores, I could see that I was beating them for GPU power (with the exception of 1 670 grrrr) and I was trailing well behind with the I7's in the physics scores.

If you do a breakdown though, you can clearly see the scores for the CPU and GPU. We have 1 guy who has a 680 running with an E chip and you can clearly see his GPU is scoring like it should but his E chip is scoring very low.

http://www.3dmark.com/3dm11/3930922

I would be saying from that bench, he is seeing ~ 40% of a bottleneck. I know and fully appreciate that this is not 'set in stone' because games like RTS/Sims/FPS all work differently. One thing it does do though is give us a rough idea of a bottleneck, which is what I said in my last post.

Robes · 9 Aug 2012 at 11:46

gregster said:
You fail to miss the OP's point!

It is a good reference to somebody who may just want to upgrade a GPU for the time being and at least they can see what kind of bottleneck will happen.

He didn't fail to miss. He may have failed to see OR missed, but he certainly didn't fail to miss the point.

Although it's nice to have something to look at, the benchmarks displayed do not tell that much of a story. As others have said games benchmarks would have been a better test, as their not specifically designed to highlight the differences between CPUs, unlike out and out benchmarks.

Something like this is a better example of CPU scaling. But I understand that not everyone wants to spend hours benchmarking.

gregster said:
http://www.3dmark.com/3dm11/3930922

I would be saying from that bench, he is seeing ~ 40% of a bottleneck. I know and fully appreciate that this is not 'set in stone' because games like RTS/Sims/FPS all work differently. One thing it does do though is give us a rough idea of a bottleneck, which is what I said in my last post.

Sorry fella, but that's not a bottleneck. A bottleneck would be if his CPU's performance was dragging down his GPU. His GPU score is pretty much what you'd expect, which shows that his CPU is doing enough to keep the GPU at full speed.

jaybee · 9 Aug 2012 at 11:57

MeMeMeMe said:
I'd have preferred seeing actual gaming benchmarks.

Major request here to OP.

OP would you be so kind enough to run some modern day gaming benchmarks and fps scores on the q6600 vs the newer i7 CPU you have on the same card? Would it be possible to also overclock the q6600 up a bit more? Most people run them at 3.2ghz minimum, and many upto nearer 4ghz.

andybird123 · 9 Aug 2012 at 12:12

Robes said:
Sorry fella, but that's not a bottleneck. A bottleneck would be if his CPU's performance was dragging down his GPU. His GPU score is pretty much what you'd expect, which shows that his CPU is doing enough to keep the GPU at full speed.

+1... it got right on my goat last year when people kept telling me my Q6600 would be bottlenecking my GTX580... but in BF3 etc. I was seeing 95-99% utilisation on the 580 and FPS scores within 1-2% of what they should be for a 580

I define bottleneck as CPU at 100% and GPU <99% with FPS scores suffering as a result (low res results don't matter either - the res you game at does)

so yes it was bottlenecking, but at 2% it wasn't worth spending £300 for 2%
a 670 on the other hand would have been a big bottleneck and given no gains over the 580, so a new CPU was needed to drive it

the problem with all these scores is that 3D mark places big emphasis on CPU - shame the 7850 heaven run couldn't complete properly as an actual FPS score of some description would have been useful

Gregster · 9 Aug 2012 at 12:21

Robes said:
He didn't fail to miss. He may have failed to see OR missed, but he certainly didn't fail to miss the point.

Although it's nice to have something to look at, the benchmarks displayed do not tell that much of a story. As others have said games benchmarks would have been a better test, as their not specifically designed to highlight the differences between CPUs, unlike out and out benchmarks.

Something like this is a better example of CPU scaling. But I understand that not everyone wants to spend hours benchmarking.

Sorry fella, but that's not a bottleneck. A bottleneck would be if his CPU's performance was dragging down his GPU. His GPU score is pretty much what you'd expect, which shows that his CPU is doing enough to keep the GPU at full speed.

Good link and again if you read my post properly I state

I know and fully appreciate that this is not 'set in stone' because games like RTS/Sims/FPS all work differently. One thing it does do though is give us a rough idea of a bottleneck, which is what I said in my last post.

Also how is his score not a bottleneck? I score P8648 and he scores P5458. Please explain. The problem is the amount of variables involved with different games, you would never be able to define a particular bottleneck. Play BF3 on a Q chip with a 7950 and then play FSX with the same setup.