Review Asus ROG Zephyrus G14 Review: AMD Ryzen 9 4900HS Tested

[lobz]

I didn't know Piednol worked in Haifa.

The funniest thing is, which actually tells the most about his character: tuned at gaming 🤣🤣🤣
NOBODY EVER tuned ANYTHING for gaming at Intel when designing an uarch.

 

[Panino Manino]

The funniest thing is, which actually tells the most about his character: tuned at gaming 🤣🤣🤣
NOBODY EVER tuned ANYTHING for gaming at Intel when designing an uarch.

He didn't said that.
He said that skylake was "tuned", but without specifying for what, and one of the results is the performance in gaming.

 

[lobz]

He didn't said that.
He said that skylake was "tuned", but without specifying for what, and one of the results is the performance in gaming.

You either speak even worse English than him, or you just follow him for so long that you can read his mind. What he wrote does not mean what you state now, at least not in English.

 

[Panino Manino]

You either speak even worse English than him, or you just follow him for so long that you can read his mind. What he wrote does not mean what you state now, at least not in English.

Yes, my english is bad, but still.
He was bragging that he was part of what made Skylake good, that he helped tweak that thing to extract the maximum performance the arch could deliver. The performance "at" gaming is a consequence of this, not that Skylake was made thinking about how to have the best performance "in" gaming.

 

[DrMrLordX]

It's best not to take Francois too seriously in any case.

 

[Panino Manino]

It's best not to take Francois too seriously in any case.

You take David Kanter seriously?

https://twitter.com/i/web/status/1250835688510984193

@FPiednoel Isn’t AMD better right now in IPC? Tweet added by Fr3dweiser @fr3dweiser | Twuko

Tweet added by Fr3dweiser @fr3dweiser @FPiednoel Isn’t AMD better right now in IPC? | Twuko

www.twipu.com

 

[AtenRa]

You take David Kanter seriously?

https://twitter.com/i/web/status/1250835688510984193

@FPiednoel Isn’t AMD better right now in IPC? Tweet added by Fr3dweiser @fr3dweiser | Twuko

Tweet added by Fr3dweiser @fr3dweiser @FPiednoel Isn’t AMD better right now in IPC? | Twuko

www.twipu.com

Well , IPC is application depended. ZEN 2 may have higher IPC in Floating Point in Cinebench but that doesnt actually mean it has the higher IPC in general.
I dont know it may have the higher IPC but taking only Cinebench to showcase this is not the complete picture.
I would like to see a review with 10-15 applications having both Integer and Float benchmarks and then we can come to a conclusion who has the higher IPC.

 

[DrMrLordX]

@Panino Manino

Maybe. But anyone can be wrong. Also Kanter may just not want to be banned from Francois' twitter feed.

 

[DrMrLordX]

Well , IPC is application depended.

Remember we're talking Renoir vs. Intel mobile chips specifically. You can't just compare Zen2 to Coffeelake on desktop and come to a conclusion. In mobile, you've got to normalize for power (I guess?) for things to be meaningful. Unless it's okay for Intel to win benchmarks by burning up 2x more package power.

 

[Hitman928]

You take David Kanter seriously?

https://twitter.com/i/web/status/1250835688510984193

@FPiednoel Isn’t AMD better right now in IPC? Tweet added by Fr3dweiser @fr3dweiser | Twuko

Tweet added by Fr3dweiser @fr3dweiser @FPiednoel Isn’t AMD better right now in IPC? | Twuko

www.twipu.com

I have a ton of respect for David Kanter and his knowledge of digital IP design far outweighs mine. With that said, appeal to authority doesn't go very far with me so without an argument with data to back it up, a simple no response is meaningless to me.

Now, from what I've seen, he is correct in terms of Icelake, but Icelake also doesn't have any skus with more than 4 cores, doesn't scale well in frequency compared to AMD or intel's own 14 nm, and doesn't seem to offer any significant power savings to speak of compared to intel 14 nm, so it's hard to mark that one in the win column.

 

[uzzi38]

I have a ton of respect for David Kanter and his knowledge of digital IP design far outweighs mine. With that said, appeal to authority doesn't go very far with me so without an argument with data to back it up, a simple no response is meaningless to me.

Now, from what I've seen, he is correct in terms of Icelake, but Icelake also doesn't have any skus with more than 4 cores, doesn't scale well in frequency compared to AMD or intel's own 14 nm, and doesn't seem to offer any significant power savings to speak of compared to intel 14 nm, so it's hard to mark that one in the win column.

The question was about IPC though, so he's right. Ice Lake is ahead in IPC.

It may be littered with other problems, but if you're gonna talk about IPC, then currently in x86 Intel is still technically in the lead.

We'll see if that holds up later this year though.

 

[AtenRa]

Remember we're talking Renoir vs. Intel mobile chips specifically. You can't just compare Zen2 to Coffeelake on desktop and come to a conclusion. In mobile, you've got to normalize for power (I guess?) for things to be meaningful. Unless it's okay for Intel to win benchmarks by burning up 2x more package power.

IPC will be almost the same in desktop vs mobile.

perf/watt is a different measurement, it doesnt have an impact in IPC performance.

 

[AtenRa]

The question was about IPC though, so he's right. Ice Lake is ahead in IPC.

It may be littered with other problems, but if you're gonna talk about IPC, then currently in x86 Intel is still technically in the lead.

We'll see if that holds up later this year though.

At what applications ??

 

[DrMrLordX]

IPC will be almost the same in desktop vs mobile.

Not necessarily! Renoir has a reworked die design. Cache latency and memory latency may be different.

 

[uzzi38]

At what applications ??

On average of a multiple applications. I dunno, if you want an industry standard or something use Spec results, they'll tell you the same thing. Sunny Cove is ahead of Zen 2 and Skylake.

 

[Hitman928]

The question was about IPC though, so he's right. Ice Lake is ahead in IPC.

It may be littered with other problems, but if you're gonna talk about IPC, then currently in x86 Intel is still technically in the lead.

We'll see if that holds up later this year though.

My point is that when your perf/clock lead is with a chip that is mobile only with a limited number of SKUs, it's not exactly a win. AMD has the perf/clock lead in desktop, server, and mobile above 4 cores. Intel has a lead in mobile 4 cores and below where it still loses in overall performance and efficiency. So who has the "IPC lead"? It's not a straight forward decision at this particular point in time.

 

[moinmoin]

Now, from what I've seen, he is correct in terms of Icelake, but Icelake also doesn't have any skus with more than 4 cores, doesn't scale well in frequency compared to AMD or intel's own 14 nm

Comparing Icelake with Intel's 14nm CPUs or Zen 2 in desktop performance right now is like comparing Apple's A13 with Intel's 14nm CPUs or Zen 2 in desktop performance! lol

 

[AtenRa]

Not necessarily! Renoir has a reworked die design. Cache latency and memory latency may be different.

Yea ok there will be a small difference.

 

[AtenRa]

On average of a multiple applications. I dunno, if you want an industry standard or something use Spec results, they'll tell you the same thing. Sunny Cove is ahead of Zen 2 and Skylake.

Yes ok, what are those applications ?? do you have a link with results from those applications ??

 

[tamz_msc]

Yes ok, what are those applications ?? do you have a link with results from those applications ??

SPEC itself is a bunch of different "applications".

 

[AtenRa]

SPEC itself is a bunch of different "applications".

We have links of Spec int and Spec float in a single thread at a given clock speed for ZEN 2 vs Ice Lake ??

 

[tamz_msc]

We have links of Spec int and Spec float in a single thread at a given clock speed for ZEN 2 vs Ice Lake ??

Not single thread and not at fixed clock speeds as Andrei has repeatedly said that its not relevant, but here you go:

The Ice Lake Benchmark Preview: Inside Intel's 10nm

www.anandtech.com

As you can see from the final graph in that page Ice Lake is ~10% ahead in PPC vs Zen 2 in int and ~13% ahead in fp.

 

[AtenRa]

Not single thread and not at fixed clock speeds as Andrei has repeatedly said that its not relevant, but here you go:

The Ice Lake Benchmark Preview: Inside Intel's 10nm

www.anandtech.com

As you can see from the final graph in that page Ice Lake is ~10% ahead in PPC vs Zen 2 in int and ~13% ahead in fp.

nice thx

this is not single thread ??

 

[tamz_msc]

this is not single thread ??

Yes, it is single thread as in its the result for a single instance of the rate benchmark(hence rate_1).

 

[AtenRa]

Yes, it is single thread as in its the result for a single instance of the rate benchmark(hence rate_1).

Nice, we have a winner then