iMac M1: Benchmarking

[jido]

Should we expect higher performance from the M1 in the new iMac, considering it's the same chip but it has two fans and a 145W power supply?

When will we see the first benchmarks?

 

[CMMChris]

No. M1 has never been thermally limited except for the MacBook Air. Also, a stronger power supply doesn't magically make the M1 faster. The iMacs will perform identical to a M1 Mac Mini or MacBook Pro.

The hardware inside the new iMac will consume roughly 80 to 90 watts with full screen brightness. Likely Apple went for a larger power supply so they can re-use it later when more beefy processors are available. Also, having a stronger power supply than needed can be good from an efficiency perspective.

 

[xraydoc]

Also, having a stronger power supply than needed can be good from an efficiency perspective.

...and extra for bus-powered peripherals.

I'm pretty confident the iMac M1 will perform exactly the same as the Mac mini M1, otherwise Apple would have made a big deal about it.

 

[leman]

It is possible that the iMac GPU is clocked higher. Apple made some performance claims during the event that imply higher scores that we know from M1.

 

[JohnnyGo]

It is possible that the iMac GPU is clocked higher. Apple made some performance claims during the event that imply higher scores that we know from M1.

That’s good news. Can you be more specific? I didn’t pay close enough attention during the event.

 

[Spindel]

Should we expect higher performance from the M1 in the new iMac, considering it's the same chip but it has two fans and a 145W power supply?

When will we see the first benchmarks?

I expect it to preform on par with the M1 Mini that has a 150 W PSU, even if the mini only has one fan (for the record my M1 minis fan has never left the 1690-1710 rpm range ever under normal use).

 

[leman]

That’s good news. Can you be more specific? I didn’t pay close enough attention during the event.

They mentioned that it's up to 50% faster than the Pro Vega 20, which the M1 we know clearly is not...

 

[CMMChris]

They did not. The 50% faster GPU claim was referring to the 21 inch iMacs it replaces. The Pro Vega 20 never was an option in those. Radeon 560X was the maximum you could get. And the M1 GPU is indeed roughly 50% faster.

 

[jinnyman]

Increased thermal envelop should provide more room for clock speed increase. I hope M1 in iMac is better performing than ones in portables. But lack of heat pipe is worrisome. Others may be correct that M1 in iMac may be identical to ones in portable (except air which throttles down when stressed).

 

[EzisAA]

They did not. The 50% faster GPU claim was referring to the 21 inch iMacs it replaces. The Pro Vega 20 never was an option in those. Radeon 560X was the maximum you could get. And the M1 GPU is indeed roughly 50% faster.

 

[CMMChris]

Oh, that's new to me. In that case Apple's marketing claim is probably bull****. I highly doubt they increased the GPU clock.

 

[jeanlain]

It is possible that the iMac GPU is clocked higher. Apple made some performance claims during the event that imply higher scores that we know from M1.

Performance claims including "up to" are vague and unverifiable. I personally think that the M1 is the same as in the other M1 Macs that have 8 GPU cores.
And this time, Apple has not posted a single benchmark result comparing the 24" iMac to previous models.

 

[jeanlain]

They did not. The 50% faster GPU claim was referring to the 21 inch iMacs it replaces. The Pro Vega 20 never was an option in those. Radeon 560X was the maximum you could get.

Nope, the Vega 20 was an option for 21.5" iMacs and wasn't available for 27" iMacs.

EDIT: already said.

 

[leman]

They did not. The 50% faster GPU claim was referring to the 21 inch iMacs it replaces.

That's what I was saying too, but someone pointed out to me that I was wrong. So I went back and rewatch that part, they explicitly say 100% faster than the base iMac and up to 50% faster than Vega Pro 20

Performance claims including "up to" are vague and unverifiable. I personally think that the M1 is the same as in the other M1 Macs that have 8 GPU cores.

Oh, we fully agree. That's why it would be interesting to actually see the how it performs.

 

[CMMChris]

Does anyone in here have proper benchmark results of a 21 inch iMac with the Vega 20? I mostly find results of the MacBook Pro's Vega 20 which likely is not the same (clocks).

 

[thenewperson]

Does anyone in here have proper benchmark results of a 21 inch iMac with the Vega 20? I mostly find results of the MacBook Pro's Vega 20 which likely is not the same (clocks).

The Metal benchmarks of the Pro Vega 20 on Geekbench are for the MBPs?

 

[CMMChris]

There are some of the iMac but I said "proper benchmarks". Geekbench is just one part of the story.

 

[quarkysg]

And this time, Apple has not posted a single benchmark result comparing the 24" iMac to previous models.

Probably due to the backlash from their last graphs? Apple likely will just let the YouTubers do it for them.

 

[jeanlain]

Probably due to the backlash from their last graphs? Apple likely will just let the YouTubers do it for them.

The baclkash was about the graph comparing the M1 to some unspecified PC CPU, which lacks units on axes, and vague claims such as "3 times faster". The graphs comparing M1 Macs to previous models were fine. These specified the app used, at least.

 

[CheesePuff]

The footnote clearly states on Apple.com this was GPU testing against the 560X and not the Vega 20:

Testing conducted by Apple in March 2021 using preproduction 24-inch iMac systems with Apple M1 chip and 8-core GPU, as well as production 3.0GHz 6-core Intel Core i5-based 21.5-inch iMac systems with Radeon Pro 560X with 4GB of GDDR5 memory, all configured with 8GB RAM and 256GB SSD.

 

[leman]

The footnote clearly states on Apple.com this was GPU testing against the 560X and not the Vega 20:

Since people seem to keep missing this, here is a link with timestamp

So: 100% faster than 560X, 50% faster than Pro Vega 20.

Anyway, these are just claims. We need some proper tests. It's a shame that there are no good GPU benchmark tools for macOS...

 

[Significant1]

They mentioned that it's up to 50% faster than the Pro Vega 20, which the M1 we know clearly is not...

It seems pretty much on par,but if you cherrypick you can end up saying 267% faster. But tests where it is a bit slower.
https://www.notebookcheck.net/M1-8-Core-GPU-vs-Radeon-Pro-Vega-20_10552_9391.247598.0.html
(not so many results from m1, so you have to scroll down quite a bit)

 

[leman]

It seems pretty much on par,but if you cherrypick you can end up saying 267% faster. But tests where it is a bit slower.
https://www.notebookcheck.net/M1-8-Core-GPU-vs-Radeon-Pro-Vega-20_10552_9391.247598.0.html
(not so many results from m1, so you have to scroll down quite a bit)

Exactly. Based on what I've seen so far, the M1 is at best similar to Pro Vega 20. It might be a bit faster in simple game benchmarks (TBDR showing it's advantages most likely), but slower overall on more complex things. So if Apple claims that the iMac has 50% faster graphics than HBM2 Vega 12, we need to ask ourselves whether it means that the iMac GPU is clocked higher than the "regular" M1 or whether it's just some BS marketing.

 

[Significant1]

Exactly. Based on what I've seen so far, the M1 is at best similar to Pro Vega 20. It might be a bit faster in simple game benchmarks (TBDR showing it's advantages most likely), but slower overall on more complex things. So if Apple claims that the iMac has 50% faster graphics than HBM2 Vega 12, we need to ask ourselves whether it means that the iMac GPU is clocked higher than the "regular" M1 or whether it's just some BS marketing.

They say UP TO and it already is/can be. Lets see when the reviews start, but I think it will be clocked exectly the same and performance on par with Mac Mini M1.

 

[JohnnyGo]

They say UP TO and it already is/can be. Lets see when the reviews start, but I think it will be clocked exectly the same and performance on par with Mac Mini M1.

I agree. All M1 seem to have similar clock speeds.

The question is whether the iPad Pro version will be under clocked to save battery. If not that’s going to blow up all benchmarks (in terms of iOS devices)