Friday, May 17th 2019

AMD Ryzen "Picasso" APU Clock Speeds Revealed

May 17th, 2019 02:58 Discuss (42 Comments)

AMD is giving finishing touches to its Ryzen 3000 "Picasso" family of APUs, and Thai PC enthusiast TUM_APISAK has details on their CPU clock speeds. The Ryzen 3 3200G comes with 3.60 GHz nominal clock-speed and 4.00 GHz maximum Precision Boost frequency; while the Ryzen 5(?) 3400G ships with 3.70 GHz clock speeds along with 4.20 GHz max Precision Boost. The "Picasso" silicon is an optical shrink of the 14 nm "Raven Ridge" silicon to the 12 nm FinFET process at GlobalFoundries, the same one on which AMD builds "Pinnacle Ridge" and "Polaris 30."

Besides the shrink to 12 nm, "Picasso" features upgraded "Zen+" CPU cores that have improved Precision Boost algorithm and faster on-die caches, which contribute to a roughly 3% increase in IPC on "Pinnacle Ridge," but significantly improved multi-threaded performance compared to 1st generation Ryzen. Clock speeds of both the CPU cores and the integrated "Vega" iGPU are expected to increase. Both the 3200G and 3400G see a 100 MHz increase in nominal clock-speed, and 300 MHz increase in boost clocks, over the chips they succeed, the 2200G and 2400G, respectively. The iGPU is rumored to receive a similar 100-200 MHz increase in engine clock.

Source: TUM_Apisak (Twitter)

Add your own comment

42 Comments on AMD Ryzen "Picasso" APU Clock Speeds Revealed

#26

Valantar

bugBrought to you by the makers of Rebrandeon.

Invest in a brand/nomenclature. Promote it. Try to to shove as many unrelated products under the same umbrella.

So Raven Ridge is somehow not Ryzen? Yeah, you're going a bit far there.

#27

Darmok N Jalad

jmcslobThe first AM4 APU'S were the same as the last generation from AM3+...7870k.
The second was 1st generation Ryzen and was released just prior to the second generation of Ryzen...

I don't think the iGPU has really gotten all that much better... The only thing that seems to really progress is the CPU cores.

Maybe they'll really take off with DDR5

I wonder if socket AM5 will offer more memory channels. If so, 4 channel would really make a difference, or maybe someday some HBM love.

#28

GoldenX

bugBrought to you by the makers of Rebrandeon.

Invest in a brand/nomenclature. Promote it. Try to to shove as many unrelated products under the same umbrella.

Skylake wants that meme back.

#29

Valantar

Darmok N JaladI wonder if socket AM5 will offer more memory channels. If so, 4 channel would really make a difference, or maybe someday some HBM love.

Extremely unlikely, as more memory channels will mean a much larger socket and much more expensive motherboards (due to the extra layers needed to route ~twice as many DRAM traces through the board). Besides, DDR5 is going to be in place by the time AM4 is replaced, with far higher bandwidth than DDR4 (at the cost of some latency, but not much). From what I've read, DDR5-5000 is going to be on the low end.

#30

GoldenX

ValantarExtremely unlikely, as more memory channels will mean a much larger socket and much more expensive motherboards (due to the extra layers needed to route ~twice as many DRAM traces through the board). Besides, DDR5 is going to be in place by the time AM4 is replaced, with far higher bandwidth than DDR4 (at the cost of some latency, but not much). From what I've read, DDR5-5000 is going to be on the low end.

That sounds like good IGP performance.

#31

Valantar

GoldenXThat sounds like good IGP performance.

Yep, I'm pretty hopeful that we'll see some really good APUs on AM5(?).

#32

Nihilus

Pointless release - only to confuse customers with the 3000 moniker. Get a 2400g for under $150 with 2 free games, then have some fun delidding and matching the clocks on these new APUs. The memory controller on the Raven Ridge are respectable as I hit 3466 cas 14 on a budget board.

#33

Valantar

NihilusPointless release - only to confuse customers with the 3000 moniker. Get a 2400g for under $150 with 2 free games, then have some fun delidding and matching the clocks on these new APUs. The memory controller on the Raven Ridge are respectable as I hit 3466 cas 14 on a budget board.

It's an iterative spec bump, so, sure, you could argue it's pointless, or you could argue that getting slightly higher clocks at slightly lower power draw for the same price is a nice bonus. Depends how you look at it. I doubt there'll be much of an IPC bump over RR, but 1-2% is likely.

#34

Nihilus

Darmok N JaladI wonder if socket AM5 will offer more memory channels. If so, 4 channel would really make a difference, or maybe someday some HBM love.

Kinda wish they would bring back triple channel memory. DDR5 in triple channel could feed one hell of an iGPU.

#35

Unregistered

I don't see why AMD can't do a separate IGP IMC with a 4gb HBM stack the same way I think the actual 3000 series will be done.. on infinity fabric... Maybe they will on AM5

#36

GoldenX

jmcslobI don't see why AMD can't do a separate IGP IMC with a 4gb HBM stack the same way I think the actual 3000 series will be done.. on infinity fabric... Maybe they will on AM5

Price.

#37

Valantar

jmcslobI don't see why AMD can't do a separate IGP IMC with a 4gb HBM stack the same way I think the actual 3000 series will be done.. on infinity fabric... Maybe they will on AM5

There's no doubt that they can - after all they can put whatever they want into the I/O die, including multiple memory controllers. The real-world issues are cost and packaging, though. First off this would be very, very expensive, and might require an interposer, adding even more cost. Secondly, there's no way they could fit a HBM die and a GPU chiplet on an AM4 (Matisse) package, at least not without dramatically shrinking the I/O die. Which would mean cutting features or an expensive node shrink, if not both.

#38

bug

jmcslobI don't see why AMD can't do a separate IGP IMC with a 4gb HBM stack the same way I think the actual 3000 series will be done.. on infinity fabric... Maybe they will on AM5

Intel did that when they packaged Vega graphics. It turned out that chip can't decode video on the Vega because of some arcane HDCP requirements. So there are constraints other than price, too.

#39

Valantar

bugIntel did that when they packaged Vega graphics. It turned out that chip can't decode video on the Vega because of some arcane HDCP requirements. So there are constraints other than price, too.

While AMD's track record on HTPC driver support isn't exactly stellar, I sincerely doubt they'd have any such issues with an MCM APU. The HDCP issues with Kaby Lake-G was due to the AMD GPU being connected over PCIe, qualifying as a dedicated GPU by the HDCP consortium's standards, heightening security/DRM requirements (as the decoded video signal passes through an unencrypted internal bus, IIRC). An APU would use IF, not PCIe for its internal connectivity. What would actually be needed would be a driver/software/firmware push from AMD to get the newest certifications for UHD Blu-ray playback and similar modern high-end HTPC features. I'm kind of baffled that these haven't already arrived, but I suppose it's such a small niche of users that it might not be worth prioritizing. I mean, I am among the relative minority who has a dedicated HTPC, but I still don't have a 4k TV, and prefer to play Blu-rays from my PS3 (which is almost exclusively used for that pupose these days).

#40

bug

ValantarWhile AMD's track record on HTPC driver support isn't exactly stellar, I sincerely doubt they'd have any such issues with an MCM APU. The HDCP issues with Kaby Lake-G was due to the AMD GPU being connected over PCIe, qualifying as a dedicated GPU by the HDCP consortium's standards, heightening security/DRM requirements (as the decoded video signal passes through an unencrypted internal bus, IIRC). An APU would use IF, not PCIe for its internal connectivity. What would actually be needed would be a driver/software/firmware push from AMD to get the newest certifications for UHD Blu-ray playback and similar modern high-end HTPC features. I'm kind of baffled that these haven't already arrived, but I suppose it's such a small niche of users that it might not be worth prioritizing. I mean, I am among the relative minority who has a dedicated HTPC, but I still don't have a 4k TV, and prefer to play Blu-rays from my PS3 (which is almost exclusively used for that pupose these days).

I didn't say it wasn't possible to do. Just that there are other aspects involved ;)
Realistically speaking though, when you want to put all/most of the components on a GPU sized PCB onto a CPU die, something's got to give.

#41

Valantar

bugI didn't say it wasn't possible to do. Just that there are other aspects involved ;)
Realistically speaking though, when you want to put all/most of the components on a GPU sized PCB onto a CPU die, something's got to give.

Oh, absolutely. Die size and power first of all - most everything else is shared, after all. Power delivery? Same stuff, just wire it to fit. Display pipeline? Already on motherboards. PCIe? Unnecessary. The main challenge would actually be cooling and power delivery, as cooling a 250W package in a CPU socket would require an extremely hefty cooler, while powering it would mean these chips only work on a handful of motherboards. And so on. Of course heat density would also be an issue with a power hungry GPU and a CPU beneath the same IHS. Still, there ought to be a future in mid-range 100-120W APUs as long as they can give them suitable memory.

#42

bug

ValantarOh, absolutely. Die size and power first of all - most everything else is shared, after all. Power delivery? Same stuff, just wire it to fit. Display pipeline? Already on motherboards. PCIe? Unnecessary. The main challenge would actually be cooling and power delivery, as cooling a 250W package in a CPU socket would require an extremely hefty cooler, while powering it would mean these chips only work on a handful of motherboards. And so on. Of course heat density would also be an issue with a power hungry GPU and a CPU beneath the same IHS. Still, there ought to be a future in mid-range 100-120W APUs as long as they can give them suitable memory.

Yeah, it's possible. Intel does Iris Pro with eDRAM, so we know it's possible. Just don't expect a fully fledged Vega56 under there ;)

Add your own comment

AMD Ryzen "Picasso" APU Clock Speeds Revealed

42 Comments on AMD Ryzen "Picasso" APU Clock Speeds Revealed

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts

AMD Ryzen "Picasso" APU Clock Speeds Revealed

Related News

42 Comments on AMD Ryzen "Picasso" APU Clock Speeds Revealed

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts