Intagrated vs. standalone GPU performance difference

[Rolanddes]

Hi guys,

When I check Intel Iris Pro Graphics 5200 vs Nvidia GeForce GT 750M from notebookcheck.net I see that they are both midrange graphics cards which do not have a lot of performance difference. Yes off course 750M is better but not the monster that I expected when I compare it to Iris.

Other than the facts that Iris is a little less powerful and it uses system memory what disadvantages does it have? What am I missing?

Thanks.

 

[maflynn]

I think the Iris Pro is probably one of the better iGPU's that intel made and it meets the needs for most consumers. Add in the fact that Apple has a horrible track record with dGPUs, I'm planning on an iGPU only MBP next time.

 

[Rolanddes]

I think the Iris Pro is probably one of the better iGPU's that intel made and it meets the needs for most consumers. Add in the fact that Apple has a horrible track record with dGPUs, I'm planning on an iGPU only MBP next time.

Thanks again maflynn

So I guess If I buy a macbook pro with retina display "without" an dGPU, it would be running games better than macbook pro early 2011 (with Ati Radeon HD6750) and even almost as good as with an eGPU (lik Geforce GT 750M)

 

[saturnotaku]

Look at actual game benchmarks to compare performance. The 750M destroys the Iris Pro in most every respect. If gaming is a consideration, you want a model with the 750M.

 

[maflynn]

Thanks again maflynn

So I guess If I buy a macbook pro with retina display "without" an eGPU, it would be running games better than macbook pro early 2011 (with Ati Radeon HD6750) and even almost as good as with an eGPU (lik Geforce GT 750M)

eGPU, what do you mean by eGPU. Are you looking to use an external GPU via thunderbolt, or are you referencing the discrete GPU (hence dGPU)?


If gaming is in the mix, then definitely go with the discrete GPU.

 

[dusk007]

You are not missing anything. The paring is a little odd. Back when the 750M was released it was only this one or a 760M would have pushed the heat limits quite a bit and probably resulted in throttling problems.
Pairing it with an IGP that is almost as fast was always strange. Nvidia has the better drivers but the Iris Pro is some respects in raw processing power actually equal. It is about 30% slower in games and such. That is not much considering that usually one expects about two times the performance from switching to a dedicated GPU.
In none gaming related stuff where the driver optimization isn't quite so much in Nvidia's favor they come out almost equal, with some better performance in OpenCL workload probably due to the extremely fast data exchange over the L3 Cache and edram Iris Pro benefits from.
Now that Apple never bothered to update to the 850M which is almost twice as fast as a 750M, one can assume they want to ditch dedicated GPUs and just want the current dGPU(750M) be slower then the next gen Iris Pro (for marketing).

 

[Rolanddes]

eGPU, what do you mean by eGPU. Are you looking to use an external GPU via thunderbolt, or are you referencing the discrete GPU (hence dGPU)?


If gaming is in the mix, then definitely go with the discrete GPU.

Yes, I meant dGPU. It was a typo and I fixed it now.

But what I wish to know is, how trivial is the performance difference between 750M and Iris when it comes to gaming. Big enough to go 750M definetaly?

----------

You are not missing anything. The paring is a little odd. Back when the 750M was released it was only this one or a 760M would have pushed the heat limits quite a bit and probably resulted in throttling problems.
Pairing it with an IGP that is almost as fast was always strange. Nvidia has the better drivers but the Iris Pro is some respects in raw processing power actually equal. It is about 30% slower in games and such. That is not much considering that usually one expects about two times the performance from switching to a dedicated GPU.
In none gaming related stuff where the driver optimization isn't quite so much in Nvidia's favor they come out almost equal, with some better performance in OpenCL workload probably due to the extremely fast data exchange over the L3 Cache and edram Iris Pro benefits from.
Now that Apple never bothered to update to the 850M which is almost twice as fast as a 750M, one can assume they want to ditch dedicated GPUs and just want the current dGPU(750M) be slower then the next gen Iris Pro (for marketing).

Thanks dusk. I will probably go with Iris Macbook pro.

 

[maflynn]

Yes, I meant dGPU. It was a typo and I fixed it now..

I didn't want to assume, because there are external GPUs

 

[Rolanddes]

I didn't want to assume, because there are external GPUs

Ok. Thanks for your help.