Is my 4770 held back by my CPU?

[CuriousMike]

Games I'm playing:
-------------------
COD MW2
Need for Speed Shift
Dragon Age
Dirt2 (Demo)

I have a (recently purchased) Radeon 4770 512MB

My CPU is an E2200 running at 2.8ghz with 4GB of RAM ( 3.25 useable.)

It's a mixed bag which games can run at my default monitor res of 1680x1050.
Dirt 2 benchmark is ~36/27 at 1680, and 38/28 @ 1280x800 at MEDIUM detail.
Since those benchmarks are so similar, does that nod towards the GPU or the CPU?

My local Frys often has AMD CPU + Mobo combos in the $80-$150 range.

The rest of the computer feels snappy enough for browsing and watching movies.
I noticed a decent bump in framerate when I upgraded from a 9600GTO to the 4770 a few weeks back.


Does my GPU match my CPU, or is one still mismatched?
How do you determine which is mismatched?

 

[happy medium]

I think it's a good match.
The Dirt 2 bench is crap.
A 4770 should handle the games you listed in med detail @ 1650x1050.
You gpu is a little slow for those games at high settings.
Your cpu should be fine.

Your bottleneck is you gpu.
You could benefit from a gtx260/4870 upgrade.

 

[Arkaign]

To be honest, you're probably better off watching FS/T section like a hawk, and hopefully scoring an E7000/E8000/Q6700 series processor eventually. If you flip that E2200 for $20 or so, then hopefully you can score a drop-in upgrade for well under $100. Any of the above overclocked to ~3.2ghz or more will be a tremendous upgrade.

Anyway, an E2200 @ 2.8ghz is not bad at all, although games do seem to prefer a higher amount of cahce.

 

[GaiaHunter]

It is a bit of both.

A Phenom II X3/X4 or even an Athlon II X3/X4, especially for dragon age, could be a decent upgrade, but then you could still use some more GPU power.

 

[CuriousMike]

Well, as an example, Frys has an X3 435 combo deal for $60.

If I knew that 435 was N% faster than my E2200, then I'd be all over it.

It's hard to get a feel for how much of an upgrade that is compared to my E2200 @2.8ghz, and if it's pointless to help the 4770.

 

[CuriousMike]

Then they have an X4 925 combo for $160; I assume that'd be a definite CPU advantage... but then I'd really be outrunning the 4770 (?)

 

[happy medium]

What kind of motherboard do you have ?

This guy has a e6300 (2.8 stock) that overclocks to 4.0 for 64$ shipped?
Sell you current cpu for 25$ and it could be a good 40$ upgrade.

http://vbforums.anandtech.com/showthread.php?t=2029259

Edit a e6300@ 3.4 (which it will easily do) would be like your cpu @3.5.
That cpu would match a 4890/gtx285.

 

[GaiaHunter]

It is hard to answer your question, most for lack of available data on the subject.

Your processor is somewhat old and has quite reduced cache - in the core 2 family cache has always been very important, especially in games.

Additionally you decreased resolution and your frame rate barely changed - that can point to CPU bottleneck (but that was only 1 game). You can further decrease resolution, overclock your GPU or reduce CPU speed and see what happens in various games.

Graphic cards though, are still the most important component regarding gaming. In this reviews Part 1: Building A Balanced Gaming PC and Part 2: Building A Balanced Gaming PC , you can see that the 4850 doesn't take advantage of the faster processors/more cores most of the time, with exception for games that are know to be multi-threaded.

In my opinion, for that resolution something like a 4870/GTX260 paired with a Athlon II X3/X4 (or a Phenon II X3/X4 if you want that extra performance achieved due to the L3 cache) would give you the best results.

As other posters said, grabbing a cheap E7000/E8000/Q6700, motherboard allowing, is quite a good option.

Test a bit more that 4770 and you CPU, by overclocking/underclocking both the CPU and GPU, change resolutions and note the effects in various games.

 

[GaiaHunter]

This guy has a e6300 (2.8 stock) that overclocks to 4.0 for 64$ shipped?
Sell you current cpu for 25$ and it could be a good 40$ upgrade.

I would take a X3 435 for $60 instead if the motherboard offered let the OP reuse his current DDR2 ram, especially considering the OP plays dragon age which should take advantage of the 3rd core.

 

[happy medium]

I would take a X3 435 for $60 instead if the motherboard offered let the OP reuse his current DDR2 ram, especially considering the OP plays dragon age which should take advantage of the 3rd core.

Agreed..

 

[CuriousMike]

My motherboard is a Gigabyte 945 ( GA-945GCM ); I'm sure I could find a faster C2D for it, but at the cost, suddenly buying one of Frys $60/$90/$129/$149 deals becomes tempting.

I'll read the balanced PC articles Gaia posted. Thanks for the input all.

 

[CuriousMike]

Graphic cards though, are still the most important component regarding gaming. In this reviews Part 1: Building A Balanced Gaming PC and Part 2: Building A Balanced Gaming PC , you can see that the 4850 doesn't take advantage of the faster processors/more cores most of the time, with exception for games that are know to be multi-threaded.

Wow, those charts are an eye-opener.

Since my 4770 is below a 4850 in terms of performance, I'll use the 4850 charts as a "close enough approximation."

What that says is: Faster CPU's do absolutely nothing for that video card. Those lines stay flat on just above every game, and every CPU ( Intel and AMD.)

You correctly state the GPU is likely more important... at least at my level.

 

[GaiaHunter]

What that says is: Faster CPU's do absolutely nothing for that video card. Those lines stay flat on just above every game, and every CPU ( Intel and AMD.)

Sincerely CuriousMike, yours is a very hard situation.

My gut feeling says: buy GPU.

But that E2200 looks really pitiful with 1 MB L2 cache.

If you had an AMD CPU, where cache basically doesn't mean that much, I would jump on "buy a GPU", no questions.

 

[AzN]

Strange I get 60fps with dirt 2 ultra quality 1920x1080 8xMSAA with the system below.

I really don't think it's the CPU. Dirt2 is not a CPU demanding game. I think there's something wrong with your computer. Was this a fresh format? Is your card running PCI-E16x? 4770 should have no problems running Dirt 2 ultra settings with 4xAA.

Tell us the rest of your computer specs including memory and what motherboard.

edit: I tried benching the demo again but @ 1280x800 with no AA. I got similar results as if I was running 1920x1080 8xMSAA. 60fps average. So it's the game itself that doesn't scale well with GPU.

So I ran the benchmark again but my CPU @ 2.5ghz to match your CPU level of performance @ 1920x1080 8xMSAA. I got 50fps average with my minimum frame rate being 36fps or so. So CPU does make some difference but not for your fps to be in the 30's with that CPU. Cache matters little although it makes "some" difference so I downclocked my CPU even further.

 

[CuriousMike]

I'm running Windows 7 64bit;
I have no idea what two sticks of 2GB PC6400 ram I have.
My mobo is listed above... cheap Gigabyte 945 based system.
Per the website, it does have PCI-15. http://tinyurl.com/ykrdatu

I don't know how your GTX260 is supposed to match up with a 4770.

I can't average 60 in any test;
I tried 1280x800 LOW quality preset and got 52/38.

I tried 1680x1050 HIGH quality preset and got 34/24
I tried 1280x800 HIGH quality preset and got 34/26
The above two tests mirror my MEDIUM numbers I originally posted; The graphics really aren't scaling up. A faster CPU would be an interesting test. Maybe I should try lowering my CPU speed (stock 2.2) and see what happens.

 

[toyota]

Strange I get 60fps with dirt 2 ultra quality 1920x1080 8xMSAA with the system below.

I really don't think it's the CPU. Dirt2 is not a CPU demanding game. I think there's something wrong with your computer. Was this a fresh format? Is your card running PCI-E16x? 4770 should have no problems running Dirt 2 ultra settings with 4xAA.

Tell us the rest of your computer specs including memory and what motherboard.

edit: I tried benching the demo again but @ 1280x800 with no AA. I got similar results as if I was running 1920x1080 8xMSAA. 60fps average. So it's the game itself that doesn't scale well with GPU.

So I ran the benchmark again but my CPU @ 2.5ghz to match your CPU level of performance @ 1920x1080 8xMSAA. I got 50fps average with my minimum frame rate being 36fps or so. So CPU does make some difference but not for your fps to be in the 30's with that CPU. Cache matters little although it makes "some" difference so I downclocked my CPU even further.

your cpu at 2.5 would be a bit faster than his at 2.8 wouldnt it? overall I thought the E2xxx cpus were barely faster than the A64 X2 cpus clock for clock. if thats true then you need your cpu at 2.1-2.2 to match his at 2.8.


EDIT: http://www.xbitlabs.com/articles/cpu/display/dualcore-shootout_4.html#sect0

 

[AzN]

your cpu at 2.5 would be a bit faster than his at 2.8 wouldnt it? overall I thought the E2xxx cpus were barely faster than the A64 X2 cpus clock for clock. if thats true then you need your cpu at 2.1-2.2 to match his at 2.8.


EDIT: http://www.xbitlabs.com/articles/cpu/display/dualcore-shootout_4.html#sect0

No it wouldn't. It's same architecture with 1 meg cache. If anything it's really neck and neck with my CPU. I've seen overclocking reviews of the 2160 that had very little effect with 1 meg cache. Perhaps my CPU is at best 100mhz faster with 1 meg cache. Considering my CPU is 200mhz or so slower than the 4meg cache core 2 duo.

So you link a review only to prove that E2200 is faster than x2 clock for clock? LOL

 

[AzN]

I'm running Windows 7 64bit;
I have no idea what two sticks of 2GB PC6400 ram I have.
My mobo is listed above... cheap Gigabyte 945 based system.
Per the website, it does have PCI-15. http://tinyurl.com/ykrdatu

I don't know how your GTX260 is supposed to match up with a 4770.

I can't average 60 in any test;
I tried 1280x800 LOW quality preset and got 52/38.

I tried 1680x1050 HIGH quality preset and got 34/24
I tried 1280x800 HIGH quality preset and got 34/26
The above two tests mirror my MEDIUM numbers I originally posted; The graphics really aren't scaling up. A faster CPU would be an interesting test. Maybe I should try lowering my CPU speed (stock 2.2) and see what happens.

I think your problem is your motherboard as well. 945 isn't exactly a core 2 duo mobo and is probably holding back some performance still you shouldn't be hovering in the 30's.

Download GPU-Z and see if your mobo is running PCI-E 16x. Sometimes a vid card is running 4x even if the mobo supports it.

 

[toyota]

No it wouldn't. It's same architecture with 1 meg cache. If anything it's really neck and neck with my CPU. I've seen overclocking reviews of the 2160 that had very little effect with 1 meg cache. Perhaps my CPU is at best 100mhz faster with 1 meg cache. Considering my CPU is 200mhz or so slower than the 4meg cache core 2 duo.

So you link a review only to prove that E2200 is faster than x2 clock for clock? LOL

well instead of giving your trademark "lol" why dont you actually look at what I said and what the review shows? I said it it would be barley faster and it is. the 2.2 E2200 is dead even with the 2.2 and 2.3 X2 cpus in some games while beating even the 2.4 and sometimes 2.6 in others. I would say the 2.2 E2200 is about like a 2.4-2.5 X2 overall. so your cpu at 2.2 or maybe 2.3 would be closer overall to his at 2.8 I would think but I guess it would really depend on the game. either way that means your cpu at 2.5 would certainly be a bit too fast for a direct comparison.

 

[AzN]

well instead of giving your trademark "lol" why dont you actually look at what I said and what the review shows? I said it it would be barley faster and it is. the 2.2 E2200 is dead even with the 2.2 and 2.3 X2 cpus in some games while beating even the 2.4 and sometimes 2.6 in others. I would say the 2.2 E2200 is about like a 2.4-2.5 X2 overall. so your cpu at 2.2 or maybe 2.3 would be closer overall to his at 2.8 I would think but I guess it would really depend on the game. either way that means your cpu at 2.5 would certainly be a bit too fast for a direct comparison.

It is you who needs to look at what you say. lol

Nice edit in your first reply...

So out of 5 games tested and 1 synthetic E2200 is faster than x2 clock for clock in all tests.

Oh BTW did you update your PhysX drivers. Your system is broken.

 

[toyota]

It is you who needs to look at what you say. lol

So out of 5 games tested and 1 synthetic E2200 is faster than x2 clock for clock in all tests.

Oh BTW did you update your PhysX drivers. Your system is broken.

nice passive aggressive comment there as usual. theres not a thing wrong with my system and I am getting the exact performance I should. as for as Batman goes, its you that is a liar as for as I am concerned. by the way I ran the same Batman demo fraps tests on a 4200 X2 system with my gtx260 and it was almost exactly what I was getting with my cpu at 1.6. I just didnt follow up because that completely proved that you were fool of shit and I didnt want to discuss it anymore. of course I guess that pc was broken too...


btw you were right in that your cpu at 2.5 is about equal to his at 2.8 according to these two games anyway. http://www.pcgameshardware.com/aid,...ache-test-Seven-CPUs-reviewed/Reviews/?page=5 unlike you I dont mind admitting if I am wrong.

 

[AzN]

nice passive aggressive comment there as usual. theres not a thing wrong with my system and I am getting the exact performance I should. as for as Batman goes, its you that is a liar as for as I am concerned. by the way I ran the same Batman tests on a 4200 X2 system with my gtx260 and it was almost exactly what I was getting with my cpu at 1.6. I just didnt follow up because that completely proved that you were fool of shit and I didnt want to discuss it anymore. of course I guess that pc was broken too...


btw you were right in that your cpu at 2.5 is about equal to his at 2.8 according to these two games anyway. http://www.pcgameshardware.com/aid,...ache-test-Seven-CPUs-reviewed/Reviews/?page=5 unlike you I dont mind admitting if I am wrong.

Of course I'm right. I've been a computer enthusiast and studying up on PC architecture before you were born. lol

You've got a lot of nerve 17 year old mr. Alabama who whines to his mammy to buy you a computer. I even posted proof with pictures and all and you call me a liar. Whatever I could care less about Mr 8600gt can run crysis @ high settings if I had a faster CPU.

Oh BTW I still think your system is broken. BFG10K even posted links showing batman had very little impact with faster CPU even on a 5970. lol...

Anyways you are always jacking threads and going off topic. Go play with your computer some more.

AZN, If you don't know off the personal attacks, you are going to get a vacation.
Markfw900
Anandtech Moderator.

 

[toyota]

Of course I'm right. I've been a computer enthusiast and studying up on PC architecture before you were born. lol

You've got a lot of nerve 17 year old mr. Alabama who whines to his mammy to buy you a computer. I even posted proof with pictures and all and you call me a liar. Whatever I could care less about Mr 8600gt can run crysis @ high settings if I had a faster CPU.

Oh BTW I still think your system is broken. BFG10K even posted links showing batman had very little impact with faster CPU even on a 5970. lol...

Anyways you are always jacking threads and going off topic. Go play with your computer some more.

nice childish reply and I am probably just as old as you or older. you always throw in your off topic passive aggressive remarks and it just shows what kind of person you really are. I tested Batman with my system on Vista and 7. I tested it on the 4200 X2 system with Vista. same results and same patterns. the 4200 X2 was a frame or two slower than my E8500 at 1.6. the results were consistent with both the physx and non physx runs. of course both systems must be broken though...

 

[AzN]

nice childish reply and I am probably just as old as you or older. you always throw in your off topic passive aggressive remarks and it just shows what kind of person you really are. I tested Batman with my system on Vista and 7. I tested it on the 4200 X2 system with Vista. same results and same patterns. the 4200 X2 was a frame or two slower than my E8500 at 1.6. the results were consistent with both the physx and non physx runs. again though both systems must be broken though...

Send a PM if you want to go childish and stop doing it here.

Anyways if you are done proving your e-penis power stop jacking threads unless you are trying to figure out the guys problem.

I doubt you are old as me. you were using HP just last year. Your first computer.

AZN, If you don't know off the personal attacks, you are going to get a vacation.
Markfw900
Anandtech Moderator.

 

[toyota]

Send a PM if you want to go childish and stop doing it here.

Anyways if you are done proving your e-penis power stop jacking threads unless you are trying to figure out the guys problem.

can I send you a PM like the childish one you sent me? oh wait now I am acting like you.


I was trying to help at first because I thought 2.5 seemed a little too fast for your cpu to compare to his. obviuosly though 2.5 is about right with your E6000 series cpu. anyway I was certainly not off topic unlike your little snide remarks.