AMD launches Ryzen Mobile 7 2700U & 5 2500U with Vega Graphics

[Yotsugi]

even with that in mind I find disappointing specially the Overwatch result, 2400 128bits is still 38GB/s.

3DMark results show it being configured with 256MB framebuffer.
Maybe this is the case?

 

[Dribble]

If they can sell these cheap enough, or get them into Apple laptops should be a lot of money to be made. The budget laptop market is a great target for AMD.

 

[SPBHM]

3DMark results show it being configured with 256MB framebuffer.
Maybe this is the case?

I don't think that's really a problem more memory is allocated dynamically when required,
perhaps the IGP is just running at a very low clock because of the TDP, also perhaps Vega is not very efficient with memory bandwidth and the penalty for sharing it with the CPU is high!?

 

[Yotsugi]

ega is not very efficient with memory bandwidth

That's totally, 146% not the case.

 

[Bouowmx]

Does it mean the 700 score is not sustainable, and 550 is the more realistic one?

Seems like the appropriate explanation.

I am still trying to digest "Mobile XFR" (MXFR). It increases Cinebench (R15) multi-thread sustained score by 23%. Why do I need this feature to do so? Given two laptops with sufficient cooling for ~50 W sustained, one with MXFR, one without, would the laptop without MXFR have lower sustained frequency than the laptop with MXFR would because MXFR is a processor feature, or would it perform the same, because MXFR is an AMD certification?

 

[krumme]

I don't think that's really a problem more memory is allocated dynamically when required,
perhaps the IGP is just running at a very low clock because of the TDP, also perhaps Vega is not very efficient with memory bandwidth and the penalty for sharing it with the CPU is high!?

Well what is actually the problem?
It wrecks the 620 solution to pieces. What more do you want it to do in 15w tdp and 14nm? Its still earth.

 

[krumme]

That's totally, 146% not the case.

Agree its a myth. And its even without use of tiled rendering as i can tell?

 

[Yotsugi]

Agree its a myth. And its even without use of tiled rendering as i can tell?

No, its most definitely doing fetch once. Just like the rest of the Vega family currently do.

 

[NTMBK]

even with that in mind I find disappointing specially the Overwatch result, 2400 128bits is still 38GB/s.

38GB/s which is also shared with a quad-core CPU.

 

[eek2121]

Raven Ridge is awesome, but Pinnacle Ridge next year is going to be THE chip to get. You guys just wait and see.

The improved boost functionality should help the single threaded IPC in desktop chips because the chip will be able to shut down 7 cores and boost the remaining single core to it's absolute limit. The new process AMD is using is going to let them hit a target of either 3.8 base or 4.0 base, but expect boost clocks to climb quite a bit higher if the voltage curve holds. Between all that and some other tweaks they are working on, I can't wait for Zen+. It's too bad I have to wait for the Threadripper variant.

PS: To the naysayers, I told you PR was going to have *at least* a 10% uplift in IPC, the improved boost functionality alone should provide that 10% along with the increase base/boost clocks...don't even get me started on the other 'surprises' AMD is cooking up...

 

[krumme]

...don't even get me started on the other 'surprises' AMD is cooking up...

How many minutes before you tell them yourself?

 

[Yotsugi]

Raven Ridge is awesome, but Pinnacle Ridge next year is going to be THE chip to get. You guys just wait and see.

The improved boost functionality should help the single threaded IPC in desktop chips because the chip will be able to shut down 7 cores and boost the remaining single core to it's absolute limit. The new process AMD is using is going to let them hit a target of either 3.8 base or 4.0 base, but expect boost clocks to climb quite a bit higher if the voltage curve holds. Between all that and some other tweaks they are working on, I can't wait for Zen+. It's too bad I have to wait for the Threadripper variant.

PS: To the naysayers, I told you PR was going to have *at least* a 10% uplift in IPC, the improved boost functionality alone should provide that 10% along with the increase base/boost clocks...don't even get me started on the other 'surprises' AMD is cooking up...

Better voltage regulation allowing higher or longer boosts has nothing to do with IPC.

 

[eek2121]

Better voltage regulation allowing higher or longer boosts has nothing to do with IPC.

True, IPC wasn't the term I was looking for, but rather, overall performance. My mistake for saying IPC, haven't had my morning coffee yet. You get the picture though.

 

[PeterScott]

Looks VERY good, but a grain a salt at pure marketing numbers. I am a little skeptical that an ultrabook can produce desktop performance levels for much time at all before thermal throttling.

I want to see real HW comparison.

I am not a big laptop user, but would love to see this fly in an SFF, or especially a Mac Mini refresh.

 

[IntelUser2000]

I want to see real HW comparison.

Agree. I myself would look at getting Raven Ridge, or successors for my desktop. I am not a multi-tasker by any means. I rarely have more than 2 tabs open and almost never two programs at once. Single thread performance is completely acceptable.

For Laptops though, we have to see how the real world battery life is. I have suspicions how it'll fare. If it does only 15-25% better than the predecessor in battery life, they won't get much traction in laptops.

Nevermind Tablets where battery life is paramount. The uncompetitive battery life was why Bobcat and Jaguar never was available in a Tablet form factor.

 

[Bouowmx]

Reading the footnotes (have to enumerate the file number in URL) for slide 18, I see that the Mobile XFR comparison is conflated with configurable power limit. In the end, I have no idea what MXFR does. Probably just an AMD certification, which I guess is helpful in easily picking out which laptops will have greater sustained performance.

 

[Roger Wilco]

Fantastic performance all around. I'm looking forward to reintroducing AMD, and introducing both Ryzen and Vega into my tech arsenal.

Many games in my steam library would run flawlessly on this thing.

 

[Ancalagon44]

You know what I just realized?

A couple of years ago, AMD released their ultra low power laptop chips - the ones with the Jaguar cores. Temash, Mullins, Beema, that sort of thing.

The ones that went into laptops were rated 15W and were generally quad core chips with a clockspeed of between 1 and 1.5Ghz. With a GPU with 128 cores.

Now look at what AMD can do with the same 15W - probably 3 times the CPU performance and something like 5 times the GPU performance. Sure, it costs more to make, but it is still a very cool example of how technology improves over time.

 

[Rifter]

Looking forward to some in depth reviews of this. So far looks great.

 

[OTG]

Reading the footnotes (have to enumerate the file number in URL) for slide 18, I see that the Mobile XFR comparison is conflated with configurable power limit. In the end, I have no idea what MXFR does. Probably just an AMD certification, which I guess is helpful in easily picking out which laptops will have greater sustained performance.

From page 3 of Anandtech's writeup, it looks like AMD will unlock XFR if a manufacturer can show that the cooling can handle 25 watts.
This might be a way to ensure the premium stuff is actually good. Maybe they'll be able to prevent somebody advertising a top-of-the-line, XFR enabled badass ultrabook, but cheap out on the cooling and ruin the experience with constant throttling.

 

[Yotsugi]

This might be a way to ensure the premium stuff is actually good. Maybe they'll be able to prevent somebody advertising a top-of-the-line, XFR enabled badass ultrabook, but cheap out on the cooling and ruin the experience with constant throttling.

Sounds too sane to be an AMD decision.

 

[Ancalagon44]

From page 3 of Anandtech's writeup, it looks like AMD will unlock XFR if a manufacturer can show that the cooling can handle 25 watts.
This might be a way to ensure the premium stuff is actually good. Maybe they'll be able to prevent somebody advertising a top-of-the-line, XFR enabled badass ultrabook, but cheap out on the cooling and ruin the experience with constant throttling.

Hopefully the requirements also include dual channel memory.

 

[TheGiant]

looking for this chip as an replacement for my surface pro 4 i5-6300U machine, very nice AMD

does anyone have info how can Intel respond to this?

 

[PeterScott]

looking for this chip as an replacement for my surface pro 4 i5-6300U machine, very nice AMD

does anyone have info how can Intel respond to this?

On the GPU side, it probably wont. CPU side, Cannon Lake is due soon. Not exactly a response, but it will be intersting to see both these new laptop CPUs put through the paces.

 

[jpiniero]

looking for this chip as an replacement for my surface pro 4 i5-6300U machine, very nice AMD

does anyone have info how can Intel respond to this?

IGP Gaming wise Intel doesn't have anything. The Coffee Lake 4+3e isn't going to be close really.