Info Ryzen 4000 Mobile Chips Unveiled at CES

[guachi]

Acer Swift 3 with Ryzen 7 4700U
AMD Ryzen 4000 Mobile APUs

AMD debuted (finally!) their Ryzen 4000 mobile chips at CES that are confusingly Ryzen 2 CPUs that are analogous with their Ryzen 3000 desktop parts.

I like what I see, especially on the power front. I know many people said that 7nm Ryzen chips had the potential to be very power efficient and if AMD's slide deck is to be believed they have succeeded. Most of the power efficiency has come from the 7nm process. As well, the 7nm process looks to be allowing AMD to cram up to 8 cores onto a laptop chip.

Do you guys think AMD has a product that will be as competitive on the laptop as the 3000 series is on the desktop? I'm thinking that the 4600 will be the best buy like the 3600 is in the desktop space. The problem in the laptop space is AMD needs design wins. At least on the desktop I don't need some company to choose for me, I can just by the chip myself.

 

[Panino Manino]

HP EliteBook 845 G7 Notebook PC Performance Results - UserBenchmark

www.userbenchmark.com

4800U looks like yet another ehh IGP, but attached to an awesome CPU.

We still trust UserBenchmark?

 

[Arkaign]

It just looks half baked to me, but with a dGPU to work with, the 4800 Ryzen is basically my 'perfect' laptop CPU. I feel and believe they can do better before long with RDNA versions, they have to be on the way, I'd be willing to pay a bet on it.

 

[naukkis]

The L2 cache was shared between cores in raven ridge vs non shared in summit ridge.
So it's 6MB vs 8.5MB of 'total cache', that's why performance was very 'adequate'.

What? L2 isn't shared between cores. Raven Ridge effective cache is 6MB, Summit ridge has 20MB effective cache as every core has private 512KiB.

 

[Arkaign]

We still trust UserBenchmark?

Not especially so, but the product vs product stack by score seems relatively accurate. It's just a lack of final reviews for this new Vega Apu so not much to go on otherwise.

 

[Thunder 57]

The problem here is that they are going to move Vega 8 from a $99 APU to a $300+ one and segment it all the way down.

Citation needed. AMD has never produced such a costly APU. On the desktop they topped out at what, $179? And no one knows what OEMs pay for laptop APU's.

 

[lobz]

Well, I guess I should wait for real world results before any firm declarations, but the UserBenchmark GPU results are something like 10% faster than the 4C/8T 1065g7 IGP. Which if that carries over to common performance, I can't find terribly exciting. I mean, 110 (or even 140!) percent of bleh is still blehhhh. For context, besides the raw numbers themselves, a 'decent' GPU scores at least 50% on their scale. The 1065 scores 15%, the 4800 Vega scores 16%. For context, this is right at DDR3 64-bit GT1030 levels, or fairly ghastly in 3d. The eight year old 28nm 7850 mid-range card is notably quicker despite it being basically terrible in features and memory for modern era. This lack of uptick in IGP despite them being possible many many years ago in consoles is kind of sad. The PS4 GPU portion from 7 years back is better, but of course it has GDDR5 and is also tied to that hideous Jaguar CPU side. I'm not asking for a 10+TF beast on PC APU, but dang, at least 2-3TF would be real nice.

I remember the hope the Hades Canyon Radeon NUC presented, and it's around what I'd consider 'bare minimum' for respectable 1080p gaming performance, and that even still dips into the 20fps range with massive stutters in AC Odyssey and Origins, though at least most of the time it's 30-50ish, and should be lockable at a console-like 30fps. And that scores 35% on the GPU benchmark, or well more than twice as fast as a GT1030 DDR3 / Vega 4800U APU does at 15-16%.

I mean sure though, I do like how basically all modern IGP is capable of competent desktop work at 4k60+, hardware h.26x, etc.

I also fully realize and hope I can adequately express this is my perspective on it. I wouldn't give one of these to my sons, it would be erratic under major titles, constantly having to tweak and lower settings to see playable performance. And I'm a definite budget king, I love setting up value options to donate.

I just think AMD has made the sane choice here instead of what they could have. Nothing to get excited about, and that showed how vaguely Lisa has spoken about it. "Oh and should I mention before I forget that these CUs have 59% higher performance?" - basically saying we can beat your latest and strongest iGPU without even really trying, whilst saving a lot of power and most importantly die area.
I completely see your point though. I guess I'm just cheering for AMD to successfully break into the mighty Intel's biggest consumer stronghold more than I do for actual iGPU tech advancement right now. - oops, busted!

 

[Kenmitch]

The products look interesting to say the least. Not really looking for another laptop at the moment myself. It's nice to see AMD's still firing on what looks like all cylinders without any sign of letting up. I guess in the end reviews of the products will paint a better picture of how much and if Intels last stand will hold up.

On another note comments from fans on both sides of the battlefield are amusing to read.

 

[lobz]

The products look interesting to say the least. Not really looking for another laptop at the moment myself. It's nice to see AMD's still firing on what looks like all cylinders without any sign of letting up. I guess in the end reviews of the products will paint a better picture of how much and if Intels last stand will hold up.

On another note comments from fans on both sides of the battlefield are amusing to read.

If you're a fan of 2 sides fighting, that's picking sides too, just doubly so. I don't even wanna imagine how bad that person would be, who's amused by your amusement!!!11

 

[Shivansps]

Citation needed. AMD has never produced such a costly APU. On the desktop they topped out at what, $179? And no one knows what OEMs pay for laptop APU's.

$149 on desktop because Picasso die tops out at 4C/11CU, the problem is that Renoir tops out at 8C/8CU, no way they are going to sell a 8C APU for less than $300, we are going to know for sure once they show the product lineup but this is common sence.

 

[Thunder 57]

$149 on desktop because Picasso die tops out at 4C/11CU, the problem is that Renoir tops out at 8C/8CU, no way they are going to sell a 8C APU for less than $300, we are going to know for sure once they show the product lineup but this is common sence.

Have to admit the 8C thing slipped my mind. Still, let's see what the lineup is before we get all upset over it. Who knows what they may do, after all, the 45W APU's don't even come with 8CU's. Higher end APU's with more cores should be a good thing. You are just bent out of shape over losing 3 CU's without knowing how the product will perform.

 

[zinfamous]

$149 on desktop because Picasso die tops out at 4C/11CU, the problem is that Renoir tops out at 8C/8CU, no way they are going to sell a 8C APU for less than $300, we are going to know for sure once they show the product lineup but this is common sence.

die size for Renoir is roughly 30% smaller, right? I wouldn't imagine that the pricing scales linearly with core count, no? twice the cores, but 30% smaller.

 

[maddie]

$149 on desktop because Picasso die tops out at 4C/11CU, the problem is that Renoir tops out at 8C/8CU, no way they are going to sell a 8C APU for less than $300, we are going to know for sure once they show the product lineup but this is common sence.

I see an 8C CPU selling right now for <$200 and a 6C for $85. It just happens to be a previous generation which, if I'm not mistaken, will be the case this year for Zen2 products. I really think you're make a mountain here out of wild fantasies. You might be right, but why not wait & see. What's the saying about a coward dying a thousand deaths.

 

[lobz]

die size for Renoir is roughly 30% smaller, right? I wouldn't imagine that the pricing scales linearly with core count, no? twice the cores, but 30% smaller.

Let's just say this: it looks to me that it's a lot more remunerative to produce and sell Renoir chips than either Picasso or the big Whiskey and the upcoming CML. The small ICLs (with relatively bigger GPUs though) could make an actual difference, if it was financially viable on its own. Unfortunately, they are produced on a node that's only half-HVM. High Volume wasting of expensive 10nm wafers and barely enough actual chips manufactured to be able to legally say, they are actually present on the market with 10nm products (please don't tell liahos!!!). Seems like that alone counts as financial viability to Intel right now, as they seem to look at financials from the wrong side of the horse: short and mid-term stock price. Meanwile Renoir could be the next big step to real stability and health in AMD's mid-term finances.

OnTrack(tm)

 

[lobz]

What's the saying about a coward dying a thousand deaths.

They didn't live long enough to tell anyone.

 

[IntelUser2000]

@Shivansps Renoir's GPU isn't slower than Vega. That's a hyperbole.

3DMark11 tests are out. 10% faster than 1065G7's Iris Plus. Seems the 30% gain is when its boosted to 25W or something.

According to the 3DMark11 score, its ~25% faster than MX250, but on an iGPU its misleading. 30% boost over mobile Vega/Iris Plus puts it on equal footing with the 25W version of MX 250 in games.

dGPUs don't have to worry about balancing power budgets between the CPU and the GPU. Oftentimes the synthetic benchmarks do better on an iGPU because usually they are quite intensive and stress only the GPU, when in real games it has to balance between both.

 

[Thunder 57]

I see an 8C CPU selling right now for <$200 and a 6C for $85. It just happens to be a previous generation which, if I'm not mistaken, will be the case this year for Zen2 products. I really think you're make a mountain here out of wild fantasies. You might be right, but why not wait & see. What's the saying about a coward dying a thousand deaths.

I wouldn't expect Renoir on desktop until Q3. By then it can be priced lower than $300 for the top chip.

Why Q3? Not a high priority for AMD, and that's when they should start to get considerably more 7nm wafers.

 

[RetroZombie]

What? L2 isn't shared between cores. Raven Ridge effective cache is 6MB, Summit ridge has 20MB effective cache as every core has private 512KiB.

Well I had difficulties in finding the information, why is it so hidden

Page 4 of Product Brief: AMD Ryzen™ Embedded V1000 Processor Family

Someone tested latency on raven ridge and they where much better, almost the same of only having just 512KB L2, but I couldn't find the article or the amd slides that stated the information.

 

[Thunder 57]

Well I had difficulties in finding the information, why is it so hidden

Page 4 of Product Brief: AMD Ryzen™ Embedded V1000 Processor Family

Someone tested latency on raven ridge and they where much better, almost the same of only having just 512KB L2, but I couldn't find the article or the amd slides that stated the information.

The only CPU's with a high L2 latency were the original Zen models. I think 17 cycles. RR, Zen+, even the first TR got it down to 12 I believe.

Those numbers are correct. Have to admit I was a little surprised at the lack of performance gain from a nice latency reduction. Makes you wonder if a 1MB L2 is in the future, as apparently the additional latency wouldn't hurt much. Die size could be a problem though.

Zen - Microarchitectures - AMD - WikiChip

Zen (family 17h) is the microarchitecture developed by AMD as a successor to both Excavator and Puma. Zen is an entirely new design, built from the ground up for optimal balance of performance and power capable of covering the entire computing spectrum from fanless notebooks to high-performance...

en.wikichip.org

 

[IntelUser2000]

Makes you wonder if a 1MB L2 is in the future, as apparently the additional latency wouldn't hurt much. Die size could be a problem though.

The issue with straight up doubling L2 is that the increase in performance due to larger cache can easily be offset by the latency increase. You may not even end up having 1% gain then.

Sure if you can keep the same, its a much easier decision.

 

[Thunder 57]

The issue with straight up doubling L2 is that the increase in performance due to larger cache can easily be offset by the latency increase. You may not even end up having 1% gain then.

Sure if you can keep the same, its a much easier decision.

True enough. Reminds me of the Prescott days where they had the 1MB and 2MB L2 versions. Here is a link to that review that shows the impact of the same CPU with different cache sizes/latency.

 

[Shivansps]

@Shivansps Renoir's GPU isn't slower than Vega. That's a hyperbole.

When i can get my hands on a 4200G, and compare it to a 3200G on stock, max oc both and on the same IGP freq and rams im going to know that. Right now as the 4200G could be as low as a 5CU SKU i wouldt bet anything on either case, the only thing i know for sure is that is not looking good due to forced segmentation.

Forget mobile, there is just too many variables to compare it, bandwidth alone already increases those numbers a lot. Check any memory scaling test on Raven or Picasso, the diference is large going from 2400 to 3200.

 

[scannall]

When i can get my hands on a 4200G, and compare it to a 3200G on stock, max oc both and on the same IGP freq and rams im going to know that. Right now as the 4200G could be as low as a 5CU SKU i wouldt bet anything on either case, the only thing i know for sure is that is not looking good due to forced segmentation.

Forget mobile, there is just too many variables to compare it.

Like all leaks from well... anywhere wait for the reviews.

 

[Shivansps]

Like all leaks from well... anywhere wait for the reviews.

I dont need to wait to reviews to know that 5-6CU will perform slower than 8CU, unless there is an arch change (is not), or it comes heavily stock overcloked(what put pressure on motherboard VRMs and limits OC gains), if they end up segmenting Vega and i dont see why they wouldt.

Im not going to support a profit driven segmentation of a previously avalible feature at the same price point, im clear? Im not going to be like one of those guys that were in favor of segmenting HT on intel i7s. AMD need to prove they actually made arch changes to make Vega faster to make up for the CU downgrade, at ANY TDP and not depending of extra bandwidth or higher freqs like they are doing on mobile.

 

[maddie]

I dont need to wait to reviews to know that 5-6CU will perform slower than 8CU, unless there is an arch change (is not), or it comes heavily stock overcloked(what put pressure on motherboard VRMs and limits OC gains), if they end up segmenting Vega and i dont see why they wouldt.

Im not going to support a profit driven segmentation of a previously avalible feature at the same price point, im clear? Im not going to be like one of those guys that were in favor of segmenting HT on intel i7s. AMD need to prove they actually made arch changes to make Vega faster to make up for the CU downgrade, at ANY TDP and not depending of extra bandwidth or higher freqs like they are doing on mobile.

What in the world is happening here.

Stock overclocked? seriously? Because a 7nm die can clock faster than a 12nm one, that means it's overclocked?

Using more power overclocked and stressing components? It's a different chip in the same power range. Why would you even make such a case?

Lastly, when have we never had segmentation?

AFAIK, AMD isn't selling CUs but performance, but what do I know. If you don't think it's worth the price, then don't buy it, simple.

 

[Thunder 57]

What in the world is happening here.

Stock overclocked? seriously? Because a 7nm die can clock faster than a 12nm one, that means it's overclocked?

Using more power overclocked and stressing components? It's a different chip in the same power range. Why would you even make such a case?

Lastly, when have we never had segmentation?

AFAIK, AMD isn't selling CUs but performance, but what do I know. If you don't think it's worth the price, then don't buy it, simple.

Seriously the silliness in here has really taken off.

Im not going to support a profit driven segmentation of a previously avalible feature at the same price point, im clear? Im not going to be like one of those guys that were in favor of segmenting HT on intel i7s. AMD need to prove they actually made arch changes to make Vega faster to make up for the CU downgrade, at ANY TDP and not depending of extra bandwidth or higher freqs like they are doing on mobile.

Are you done with your rant? Segmentation has always been a thing, artificial or not. Nice things cost money. Nicer things cost more money. And all this is speculation on unannounced parts with no pricing or performance numbers. It's gotten to be annoying, almost like throwing a fit because AMD isn't going to give you something you want. Did you stop buying Intel when they used cache, AVX, and HT for segmentation purposes? You are really being a buzzkill over this in case you haven't noticed.

Also, like I said, AMD didn't even announce an 8CU 45W renoir part. The reason was because all of the OEMs that were interested said they were pairing them with dGPU's. Maybe AMD is smart, and realizes the desktop APU models are going to be mostly, if not all, relying on the iGPU. Maybe there they leave the 8CU's on each part. Or at least a 6CU and 8CU. You're getting upset over unannounced parts. How does that mane any sense?