AMD launches Ryzen 4000 series APUs: Up to 8 cores / 16 threads at 15W

mongeese

TS Maniac
Staff member

AMD is launching seven new 4000-series processors, which despite the moniker, use the same Zen 2 technology that 3000-series desktop processors use. But don’t be too upset: AMD says the combination of their Zen 2 architecture and TSMC’s 7nm processor has doubled the performance per watt.

The processors are divided into two categories, the H-series for heavy loads like gaming or content creation, and the U-series for ultraportable laptops. The H-series can be paired with a powerful discrete GPU, while the U-series are meant to be used solo. AMD’s H-series can be directly compared with Intel’s H-series, but AMD’s U-series competes with both Intel’s G-series and U-series.

Model Cores/
Threads
Base Clock Boost Clock GPU CUs GPU Cores GPU Clock TDP
Ryzen 7 4800H 8/16 3.0 GHz 4.2 GHz 7 448 1600 MHz 45W
Ryzen 5 4600H 6/12 3.0 GHz 4.0 GHz 6 384 1500 MHz 45W

The Ryzen 7 4800H is the powerhouse of the lineup. When both are paired with an equivalent discrete GPU, AMD claims it can beat the hexa-core Intel Core i7-9750H by 39% in games, and by 46% in content creation workloads.

Model Cores/
Threads
Base Clock Boost Clock GPU CUs GPU Cores GPU Clock TDP
Ryzen 7 4800U 8/16 1.8 GHz 4.2 GHz 8 512 1750 MHz 15W
Ryzen 7 4700U 8/8 2.0 GHz 4.1 GHz 7 448 1600 MHz 15W
Ryzen 5 4600U 6/12 2.1 GHz 4.0 GHz 6 384 1500 MHz 15W
Ryzen 5 4500U 6/6 2.3 GHz 4.0 GHz 6 384 1500 MHz 15W
Ryzen 3 4300U 4/4 2.7 GHz 3.7 GHz 5 320 1400 MHz 15W

AMD is directly going after the Intel Core i7-1065G7 with the high-end U-series. Intel’s APU has a four-core, eight-thread CPU with a base clock of 2.6 GHz and a boost clock of 4.5 GHz and a 512 core GPU at 1.1 GHz. On paper, the 4800U basically dismantles the 1065G7, and AMD’s benchmark numbers (below) support that. But we’ll have to wait for benchmarks to see how the other processors compare.

"We've actually done a tremendous amount of optimization around these graphics cores... they have 59% more performance than the previous generation."

The 4000-series also makes for a rather peculiar platform. Supported memory speeds have increased from 2400 MHz to DDR4-3200 and LPDDR4-4266, which will seriously boost GPU performance in particular. On the other hand, AMD is not bringing PCIe 4.0 support to mobile, which is a shame.

The Ryzen APU 4000-series will become available in laptops this quarter, with over one hundred models from different manufacturers expected this year.

Permalink to story.

 
  • Like
Reactions: CharmsD

Dimitrios

TS Guru
Lisa is being very clever. Flood the market with AMD motherboards, CPU's & GPU's and rely on a longer cemented upgrade path and watch INTEL lose share within that long time frame.

Once the damage is done to INTEL it will be too late for them to change their act.
 

Porkous

TS Member
If those cores have an extra advantage of being tuned for even higher efficiency, and actually is like they said (why wouldn't) and offer desktop level of performance... Then we are very pleased. Wouldn't imagined this performance in a portable device. My first laptop had an atom and the performance was compared to a low power dual core (a biscuit) from 2007, if not even sooner (a smart phone would offer better fluidity sometimes).
With latest release, we kinda take a huge leap into the future, and if the price is correct, I don't see a use for desktop pc if everything is brought to mobility uses. It's pathetic what kind of performance we had in the latest years in laptops. Haha, intel offered 2/4 cores 4/8 threads. Haha! said AMD.
Freaking biscuits.
 

quadibloc

TS Addict
Intel does offer six-core mobile chips, although their 15 watt part has a clock of 1.1 GHz instead of 1.8 GHz. I think Intel could shrug off the new Threadripper, but without a response from Intel, these chips will be the only reasonable choice for a new laptop - and laptops are where Intel was still ahead. So I see this announcement as being an existential threat for Intel.
Previously, I figured Intel would soon get 10nm fixed, and go back to being on top shortly thereafter. Now I finally have to admit they're in real trouble.
 
  • Like
Reactions: JaredTheDragon

Danny101

TS Evangelist
"The 4000-series also makes for a rather peculiar platform. Supported memory speeds have increased from 2400 MHz to DDR4-3200 and LPDDR4-4266, which will seriously boost GPU performance in particular. On the other hand, AMD is not bringing PCIe 4.0 support to mobile, which is a shame."

They don't want their APU's competing too well with their desktop lineups.
 

Tom Yum

TS Rookie
"The 4000-series also makes for a rather peculiar platform. Supported memory speeds have increased from 2400 MHz to DDR4-3200 and LPDDR4-4266, which will seriously boost GPU performance in particular. On the other hand, AMD is not bringing PCIe 4.0 support to mobile, which is a shame."

They don't want their APU's competing too well with their desktop lineups.
More likely, PCIe 4 is a power guzzler and completely wasted in a laptop with limited connectivity options.
 

Evernessince

地獄らしい人間動物園
More likely, PCIe 4 is a power guzzler and completely wasted in a laptop with limited connectivity options.
It certainly is and costly to make. Best to go desktop if you need that much bandwidth.

"Supported memory speeds have increased from 2400 MHz to DDR4-3200 and LPDDR4-4266"

Will be interesting to see GPU performance with good RAM.

I'm also wondering if AMD will be offering lower clocked configurations of it's 45w parts. Would be interesting to see a lower clocked 8 core running on a 15w ultra-portable design.
 
  • Like
Reactions: Charles Olson

Irata

TS Maniac
I'm also wondering if AMD will be offering lower clocked configurations of it's 45w parts. Would be interesting to see a lower clocked 8 core running on a 15w ultra-portable design.
The U series are 15W parts and that includes eight core 16T parts.
Not sure how they‘re doing this and/or if they‘ll use more power under boost.
 

Evernessince

地獄らしい人間動物園
The U series are 15W parts and that includes eight core 16T parts.
Not sure how they‘re doing this and/or if they‘ll use more power under boost.
Intel's U series are 15w TDP, which is essentially worthless the way Intel measures it. They consume 15w at base clock and much more at boost.
 

Irata

TS Maniac
Intel's U series are 15w TDP, which is essentially worthless the way Intel measures it. They consume 15w at base clock and much more at boost.
AMD's U series is also 15W 😁

I am curious how accurate that TDP value is, so really looking forward to reviews.
 

Evernessince

地獄らしい人間動物園
AMD's U series is also 15W 😁

I am curious how accurate that TDP value is, so really looking forward to reviews.
The difference being that AMD does not calculate TDP the same way as Intel does. Making the comparison here that you did implying that they are in any way equal says to me that you are not aware of this so I will explain. You need simply look at their desktop processors as an indication of that. A 9900K easily goes over 240w yet has a TDP of 95w. Meanwhile the 3800X consumes 184w yet has a TDP of 105w. TDP is not a measure of power consumption, it's an arbitrary calculation that can be made to say whatever you want it to. Just ask GamersNexus. That said, if you are aware of how each company calculates TDP, you can get a general idea of heat output and power consumption.

If both an AMD and Intel CPU has the same TDP, I'm going to say the AMD CPU is going to consume less power, given that's been the case for every desktop processor so far.
 

Irata

TS Maniac
I am aware, but thanks for the effort Evernessince :)

This is why I am wondering just how AMD is getting their 8C16T CPU to be 15W and what the actual max power consumption is under load. This is an APU after all.
 

Evernessince

地獄らしい人間動物園
I am aware, but thanks for the effort Evernessince :)

This is why I am wondering just how AMD is getting their 8C16T CPU to be 15W and what the actual max power consumption is under load. This is an APU after all.
Binned chips for power consumption, an efficient boosting algorithm. If AMD managed to fit four more cores into the 3950X while increasing the clock by 0.1 GHz at a lower power consumption then the 3900X I can imagine they can pull off a similar feat elsewhere. The mobile parts are also going down a node as well, so they also have that. I read the interview on AnandTech and apparently the 7nm Vega turned out very efficient as well.