First Ryzen 7 iMac benchmarks?

[FIVR]

Saw this posted on a german forum. Not sure if real.... but it looks very interesting.

https://browser.geekbench.com/v4/compute/1279306

 

[Yotsugi]

Hackintosh?

 

[jpiniero]

Hackintosh?

Yeah.

 

[FIVR]

Hackintosh?

It has a prototype Vega 10 XT compute accelerator, so perhaps it is undergoing testing by Apple.


It would make sense for Apple to use the 1700 in an iMac, considering it performs about equally to the 8700k/7800X but uses half the power. Apple is very TDP-conscious in its designs. Intel on the other hand has been going the opposite direction (higher and higher TDP) with its designs.

 

[Yotsugi]

It would make sense for Apple to use the 1700 in an iMac

Assuming it is the case - it's a gigantic win for AMD.
Now they need TB3 implementation for RR refresh to get into the throttlebooks.

 

[dahorns]

It would make sense for Apple to use the 1700 in an iMac, considering it performs about equally to the 8700k/7800X but uses half the power.

Come on....

 

[Yotsugi]

Come on....

Check THG power measurements.

 

[dahorns]

Check THG power measurements.

Ryzen is very good, but i cannot find anything to substantiate the claim that it performs the same as the 8700k at half the power. Not even close. Maybe there is some niche scenario where that is accurate, but overall the 8700k uses more power, but delivers more performance in line with the power usage.

 

[Yotsugi]

but delivers more performance in line with the power usage.

In games?

 

[dahorns]

In games?

Yes? And in other things too. If you have something in particular youre talking about, point it out.

Anyway, it is possible Apple is testing things out. But hackintosh seems more likely. Apple is big on stability, which has not historically been a strength of AMD. I would also expect a change like this to make its way to financial rumors first.

 

[Yotsugi]

And in other things too.

It's ~1700 perf at 1.5-2 times the power in anything multithreaded.

Apple is big on stability

Hyperscale is even bigger on stability and yet they are adopting EPYC.
Did you make the SKL-SP FUD slide deck?

which has not historically been a strength of AMD.

Opterons want to talk to you.
Stop defending silly Intel firestarters in all-in-one systems.

 

[Qwertilot]

Intel have some vastly more energy efficient chips than the 8700K of course, and Apple have tended to use those on the iMacs. Zen would seemingly make a very good ~45w TDP chip too, so they're bound to be evaluating it.

 

[dahorns]

It's ~1700 perf at 1.5-2 times the power in anything multithreaded.

Opterons want to talk to you.
Stop defending silly Intel firestarters in all-in-one systems.

Show me what youre talking about. I am not defending anything, the claim is just unsupported by any data I have seen. It is an extraordinary claim, so provide some proof.

 

[Yotsugi]

the claim is just unsupported by any data I have seen

Are you telling me to spoonfeed you? 10/10.

Zen would seemingly make a very good ~45w TDP chip too, so they're bound to be evaluating it.

Zen is very very good unless you're pushing higher clocks while bumping into the voltage wall (and you're not going to do that in aio system).

 

[dahorns]

Are you telling me to spoonfeed you? 10/10.

What is your problem?

 

[FIVR]

Ryzen is very good, but i cannot find anything to substantiate the claim that it performs the same as the 8700k at half the power. Not even close. Maybe there is some niche scenario where that is accurate, but overall the 8700k uses more power, but delivers more performance in line with the power usage.

Apple does not overclock its intel SKUs and it prefers to keep low TDPs (lower than 8700k or 7800X). If they want to increase core counts while keeping a ~65watt TDP... there is simply no consumer option for intel. They would have to buy very expensive Xeon skus.


Nobody is arguing that the 1700 is faster than 8700k/7800x... simply that it offers similar MT performance at much lower power. Nobody is also arguing that this benchmark is proof that Apple is switching to intel. Merely it is evidence that Apple is evaluating R7 and may be switching iMac to Ryzen 7 from Skylake.

Doth protest too much.

 

[TahoeDust]

Apple does not overclock its intel SKUs and it prefers to keep low TDPs (lower than 8700k or 7800X). If they want to increase core counts while keeping a ~65watt TDP... there is simply no consumer option for intel.

What about the i7 8700? It is 6c/12t and 65w TDP. I don't even know when the "K" version is being talked about for Apple.

 

[Dayman1225]

Isn't the 1700s base freq 3ghz with it boosting upto 3.7.

 

[PeterScott]

I keep hoping the next Mini has Raven Ridge, so we can get some better GPU.

 

[FIVR]

What about the i7 8700? It is 6c/12t and 65w TDP. I don't even know when the "K" version is being talked about for Apple.

That might work but testing has shown that at similar performance and similar clocks (the boost clocks of the 8700 and 8700k are almost identical) Coffee Lake uses substantially more power than Ryzen 7 while it also has less cores and less I/O.

Apple has historically used "K" sku so it actually makes perfect sense to talk about them. See current lineup:
https://www.apple.com/imac/specs/
They also used the 4770k and the 4970K and the 6700K previously.

I think it likely that in a TDP-constrained platform like the iMac Apple would find that R7 gives similar performance to 8700/8700k at much lower power usage and, probably most importantly, at much lower ASP from the manufacturer. The 8700k and 8700 are 300-400$ processors with extremely low availability. Apple would also need lots of chips and it is very obvious that intel is (for the moment) unable to provide large number of 8700k or 8700.

Intel has only just been able to provide a few dozen Coffee Lake Skus to the biggest distributer in germany.

https://www.youtube.com/watch?v=somfMjeZrVY

Apple won't pick a processor that intel can only provide a few hundred of for the launch of such a high volume product as the iMac.

 

[TahoeDust]

Apple has historically used "K" sku so it actually makes perfect sense to talk about them. See current lineup:
https://www.apple.com/imac/specs/
They also used the 4770k and the 4970K and the 6700K previously.

Interesting, I did not know that. But, those also have a TDP of 84w, 88w, and 91w respectively. 95w is not that huge of a departure of what they are already using. I don't see anything that would make intel's 8th gen less fitting for the application than what they have been using.

I don't really care either way. Apple would be foolish to not explore and test all options.

 

[dahorns]

Apple does not overclock its intel SKUs and it prefers to keep low TDPs (lower than 8700k or 7800X). If they want to increase core counts while keeping a ~65watt TDP... there is simply no consumer option for intel. They would have to buy very expensive Xeon skus.


Nobody is arguing that the 1700 is faster than 8700k/7800x... simply that it offers similar MT performance at much lower power. Nobody is also arguing that this benchmark is proof that Apple is switching to intel. Merely it is evidence that Apple is evaluating R7 and may be switching iMac to Ryzen 7 from Skylake.

Doth protest too much.

I simply protested an unsupported and exaggerated claim, and asked for proof of the claim. I have no doubt that in purely multithreaded tests that the 1700 is more efficient. More cores at lower frequency will always have an efficiency advantage in those tests. But, the claim was blanket, that the 1700 offered the same performance at half the power of the 8700k. I honestly do not believe that claim is supported in pure multithreaded scenarios, much less in any single threaded tests. I do agree that the 8700k would be a poor choice for an all in one system. The 8700 would obviously be a superior choice, as would the 1700, due to thermal limitations. For some reason my skepticism invoked a rude response, which I believe was entirely unwarranted.

 

[jpiniero]

The iMac was just recently updated... actually most of the Mac line was updated in June. The Mini, yeah...

I'm telling you though, it's a Hackintosh. Apple, when they are ready to dump Intel, would be most likely to use their own processors.

 

[PeterScott]

It's ~1700 perf at 1.5-2 times the power in anything multithreaded.

Totally Untrue:

Blender:
8700K Stock: Performance: 23.7 minutes. Power 96 Watts
1700 OC: Performance: 27.6 minutes. Power 131 Watts

8700K is faster while using significantly LESS power.

https://www.gamersnexus.net/images/media/2017/CPUs/8700k/8700k-legacy-blender-2.78a-monkey.png
https://www.gamersnexus.net/images/media/2017/CPUs/8700k/8700k-power-draw-blender.png

 

[FIVR]

Totally Untrue:

Blender:
8700K Stock: Performance: 23.7 minutes. Power 96 Watts
1700 OC: Performance: 27.6 minutes. Power 131 Watts

8700K is faster while using significantly LESS power.

https://www.gamersnexus.net/images/media/2017/CPUs/8700k/8700k-legacy-blender-2.78a-monkey.png
https://www.gamersnexus.net/images/media/2017/CPUs/8700k/8700k-power-draw-blender.png

Been reading "how to lie with statistics" I see....


You are comparing an OC 1700 running way outside its efficiency peak to a stock 8700k. Nobody is fooled by this.

At stock clocks the 1700 is vastly more efficient than the 8700k. The latter MUST be clocked much higher to meet performance of the 1700 because it has 25% fewer cores. Intel SMT is also not as good as AMDs. Compare stock to stock... it's not even a contest.