AMD Ryzen 5 2400G and Ryzen 3 2200G APUs performance unveiled

[SPBHM]

I have to agree, those videos in random areas showing OK framerate on Witcher 3 can be very misleading

2 easy spots that tend to be GPU limited on Witcher 3

Crookback Bog
https://www.youtube.com/watch?v=FgN859FrBrU

this cutscene with the wild hunt
https://youtu.be/FgN859FrBrU?t=2m32s

Novigrad is also a good place to test I guess

wait he said console level fps ! quit different than desktop fps! , therefor TW3 desktop is not same as console.I believe 1080p Low quality on PC version ~ medium quality( or maybe a bit more) on console version.If he meant Full medium PC version yeah He's wrong.

console is a mix of settings, I think the experience overall is like playing on the PC with a 30 max lock, 1080P medium-high (overall closer to high, but some specific settings might be closer to low)

the videos I posted are older, with the last patch I think W3 was more stable on both consoles.

 

[IntelUser2000]

RX560 with double the Processors (1024 vs 512) and double the TMUs (64 vs 32) has the exactly same memory bandwidth (112GB/s) as RX 550.

Yet the RX 560 isn't 2x the performance of the RX 550. Polaris at the high end like the RX 480 is said to be limited by memory bandwidth. RX 560 is half of the full Polaris.

That means RX 560 is bandwidth limited as well. Bandwidth limitation also does not mean performance will stop scaling at all by increasing shaders and ROPs. You'll just get less out of it.

By AMD's own admission, 1650MHz clocked, DDR4-3600 2400G is 39% faster in Firestrike than the baseline configuration. My guess is the RX 550 will end up still being 10-15% faster. Not to mention you can overclock the RX 550 too.

 

[french toast]

I am not complaining about RR being on GF 14nm+, I am pointing out that GF 14nm+ does NOT exist. One way or another, it was abandoned in favor of "12nm".

Show me anything from GF specifying their 14nm+ process. AFAIK, there is no such process. It was likely a future outlook conveyed to AMD, that AMD used in roadmaps, but what emerged instead was 12nm, and AMD roadmaps were changed to reflect that.

OTOH, you can find information about GF 12nm, because that actually exists.

I can't say for certain whether GF 12nm is just a renaming of the roadmapped GF 14nm+, or if GF 14nm+ was scrapped in favor of GF 12nm, but either way GF 14nm+, does NOT exist.

Yes there is a "3rd generation" of 14nmlpp, this is what AMD called the process improvements to Rx 580...this could quite easily mean 14nm+ in AMD own parlance.

12nm could be 14nm ++...just with funky Nvidia/tsmc marketing.

As regards to my earlier claims, originally i said 2400G with really fast ram could probably do 1080p medium/MIX (meaning some settings off or on low like blur effects)..@ console FPS...meaning between low 20s to 30 FPS depending on area, novigrad and crookback bog being the worst areas... similar experience to console (Xbox one/PS4)
Alternatively you could drop the Res down to 900p medium, meaning complete Xbox one experience....not bad for an APU.

Besides this silly arguement about getting Bristol ridge or carrizo and putting a GPU on it...that's ridiculous, construction cores are attrocious, also trying to do direct igpu comparisons with Bristol ridge is not so easy, raven ridge memory controller is more than twice as effective as Bristol ridge, plus Vega likely has better compression, + HBCC?.

Atenra put a good breakdown of the specs, I believe 2400g overclocked with fast ram is a decent gaming experience, but there is nothing stopping you buying cheaper ram and sticking a descrete gpu on a 2200g...get better all-round performance and upgrade ability Vs g4560 + dgpu.

 

[Peter Watts]

Most non PC gamers have no problem with console graphics.

 

[PeterScott]

Yes there is a "3rd generation" of 14nmlpp, this is what AMD called the process improvements to Rx 580...this could quite easily mean 14nm+ in AMD own parlance.

We aren't talking about your personal interpretation, we are talking about actual product.

There's nothing called 14nm+, either in GF processes, nor in any current AMD documentation.

GF 14nm+ only existed on early AMD roadmaps, which were later updated to 12nm.

Ryzen Mobile at AMD:
https://www.amd.com/en/products/apu/amd-ryzen-7-2700u
"
CMOS
14nm
"

14nm, not 14nm+.

 

[raghu78]

It is the exact same process. Same fin pitch, same density, same metal layers, same libraries. As the process matures the fin height improves (this happens on literally every process in history). So do power characteristics.

So intel figured "well, the products is better, so why not call it '+'?". And the rest is history.

Same exact process, different name.

Now GF is following intel's lead, and some posters are frustrated because only intel is allowed to lie and get away with it.

You are absolutely wrong here. Intel 14 and 14+ have a Contacted Poly pitch (CPP) of 70nm while 14++ has a CPP of 84nm. 14++ also has taller fins and 24% higher driver current than 14 and 10% faster than 14+.

https://www.semiwiki.com/forum/content/6713-14nm-16nm-10nm-7nm-what-we-know-now.html

 

[FIVR]

You are absolutely wrong here. Intel 14 and 14+ have a Contacted Poly pitch (CPP) of 70nm while 14++ has a CPP of 84nm. 14++ also has taller fins and 24% higher driver current than 14 and 10% faster than 14+.

https://www.semiwiki.com/forum/content/6713-14nm-16nm-10nm-7nm-what-we-know-now.html

Perhaps 14++ and 14+ are different but 14 and 14+ are the exact same process. It is just marketing PR by intel. I don't even think 14++ is out yet so obviously that could be a real difference and I never said anything about 14++

So again I was not "absolutely wrong". Read my post. I said not a word about 14++ and I have seen ZERO documentation that says there is any difference between the process used to manufacture Skyalke(14) and the one used to manufacture Kaby Lake and Coffee Lake (14+)

 

[mikk]

Perhaps 14++ and 14+ are different but 14 and 14+ are the exact same process. It is just marketing PR by intel.

Factual wrong. To be honest you are absolutely clueless as usual.

This process as a quick summary has a higher fin height and larger pitch, essentially giving a less-dense set of transistors that have more room to breathe. Normally a larger pitch means more voltage required, but this is offset by the fin height and Intel says is good for another few hundred MHz for performance. The less-dense design, in theory, may also help in overclocking, however we will have to wait until January to see those results.

https://www.anandtech.com/show/1061...s-six-notebook-skus-desktop-coming-in-january

 

[FIVR]

Factual wrong. To be honest you are absolutely clueless as usual.


https://www.anandtech.com/show/1061...s-six-notebook-skus-desktop-coming-in-january

Yeah that's garbage. The fin height can change within the same process as yield and maturity of the process improves. The fin pitch (not sure what "pitch" they're talking about) is the exact same with 14 and 14+ so that's just wrong. Also, drive currents and power characteristics also change (improve) over time as the process matures, but it is the same process. At best, 14+ is just a new stepping of 14nm. It's not a new process at all. It's just marketing PR.

You don't know what you're talking about here. You're just believing whatever intel tells you, facts be damned. Even Raghu acknowledges that the density and CPP (pitch) is identical between 14 and 14+. I know people who work at this company (high level execs) and they have told me point blank it is the exact same process, just slowly improving their yields and improving the same process. The parameters are all identical.

 

[Shivansps]

A12-9800
8 CUs = 512 Processors (1,13GFLOPS at 1108MHz)
32 TMUs at 1108MHz = 35,45 GTexels/s
8 ROPs at 1108MHz = 8,86GPixels/s
up to DDR-4 2400 = 38,4GB/s Bandwidth

Ryzen 2200G (OC to 1550MHz + DDR-4 3600)
8 CUs = 512 Processors (1,58GFLOPS)
32 TMUs at 1550MHz = 49,6 GTexels/s
16 ROPs at 1550MHz = 24,8GPixels/s
DDR-4 3600 = 57,6GB/s Bandwidth

For comparison,

RX550 (1183MHz)
8 CUs = 512 Processors (1,21GFLOPS)
32 TMUs at 1183MHz = 37,86 GTexels/s
16 ROPs at 1183MHz = 18,93,8GPixels/s
GDDR-5 128Bit 7000MHz = 112GB/s Bandwidth

GT1030 (1468MHz boost)
384 Cores = (1,127GFLOPS)
24 TMUs at 1468MHz = 35,23 GTexels/s
16 ROPs at 1468MHz = 23,48,8GPixels/s
GDDR-5 64-Bit 6008MHz = 48,06GB/s Bandwidth

Ryzen 2200G at 1550MHz and DDR-3600 has higher specs than all the above, plus a new and better GPU architecture than A12-9800 and RX550. RX550 has almost double the Memory bandwidth vs the Ryzen 2200G but it doesnt use it because its severely under power for that huge bandwidth.
In paper Ryzen 2200G at 1550MHz and 3600 ram it should even be faster than GT1030.

OC+DDR4-3600? Someone is moving the goalposts here.
On paper the 2200G has to share that bandwidth with the CPU. Also the 1030 is nothing, it can barely hold its own on 720p, just like the RX550.
If it matches it, fine, but it will still get crushed by a 1050, so it really dosent worth to pay any extra for it.
I still find hard to belive it can even match the 1030, AMD IGPs are yet to reach R7 250 GDDR5 levels of performance, GT1030/RX550 is a bit far off. And those slides does not give me any hope at all, why will they show the 2400G vs the I5-8400 IGP, and then show you it can match the I5-8400 + GT1030 on firestrike? There can be only one reason for that, and you know it.

Not to mention about 100% over the A12-9800 should be at 72 font size on a slide, they did not even mention Bristol Ridge.

You know very well what is the reason for AMD hidding stuff(or being very very creative with the fight they pick) on the slides in the past.

 

[denywinarto]

DDR4-2400 should be the bottom low, except for internet cafes (non gaming ofc). DDR4-3000 should be an ideal for 2200G, not so sure for 2400G, though. AMD won't show its detailed specification until Feb, 12th.

Hmm after reading some posts here now i'm torn between 2200G + high freq Ram or 2400G + 2133/2400 mhz ram..
$69 price difference is kinda hard to swallow cause i'm gonna buy alot of them (bout 30) so i need to really think this through...
Most of my market is mid graphics games, but i'm worried at some point AAA games is gonna dominate and i'm gonna need higher specs..

 

[ao_ika_red]

Hmm after reading some posts here now i'm torn between 2200G + high freq Ram or 2400G + 2133/2400 mhz ram..
$69 price difference is kinda hard to swallow cause i'm gonna buy alot of them (bout 30) so i need to really think this through...
Most of my market is mid graphics games, but i'm worried at some point AAA games is gonna dominate and i'm gonna need higher specs..

Personally I'll choose 2200G and add 1030/1050 in a year or two.

 

[Shivansps]

Why would anybody like to be stuck with a Bristol Ridge APU with much slower Construction cores? I thought that consensus from previous discussions on the late availability for Bristol Ridge for desktop was that it was not very viable as an upgrade path if it bought at this juncture?

2200G provides the best upgrade path as of now, especially considering it's 100$ price, even if the iGPU isn't of much use, since it gives you a good quad-core CPU as well.

and I thought that consensus from previous discussions was 4/4 cores were dead for gaming, so why we are even considering a 4/4 2200G to be viable in any way now?

If anyone considers the G4560 as an alternative, the A8-9600 is also an alternative, its simple as that. And for gaming the A8-9600 paired with a 1050 will just destroy the 2200/2400G alone. The 9600 will have a way faster GPU avalible, more ram avalible (6 vs 8), and its cores are not going to be fighting the gpu for memory bandwidth, so even if they are slower cores, it worth it to have faster perf now, and you can upgrade later to a Ryzen CPU. Yes, a 2200G+1050 its going to be better than A8-9600+1050. But it is an alternative.

 

[wahdangun]

Hmm after reading some posts here now i'm torn between 2200G + high freq Ram or 2400G + 2133/2400 mhz ram..
$69 price difference is kinda hard to swallow cause i'm gonna buy alot of them (bout 30) so i need to really think this through...
Most of my market is mid graphics games, but i'm worried at some point AAA games is gonna dominate and i'm gonna need higher specs..

if you gonna use it in internet cafe then its better of with high frequency ram, because you can get good game performance now and then when you need more performance just sell the cpu and maybe by then there are zen2 apu based.

beside in indonesia second hand cpu sometimes still retain its resale value.

 

[ao_ika_red]

beside in indonesia second hand cpu sometimes still retain its resale value.

Couldn't agree more. Even c2d era cpu still has its value today. Unbelievable.

 

[frozentundra123456]

and I thought that consensus from previous discussions was 4/4 cores were dead for gaming, so why we are even considering a 4/4 2200G to be viable in any way now?

If anyone considers the G4560 as an alternative, the A8-9600 is also an alternative, its simple as that. And for gaming the A8-9600 paired with a 1050 will just destroy the 2200/2400G alone. The 9600 will have a way faster GPU avalible, more ram avalible (6 vs 8), and its cores are not going to be fighting the gpu for memory bandwidth, so even if they are slower cores, it worth it to have faster perf now, and you can upgrade later to a Ryzen CPU. Yes, a 2200G+1050 its going to be better than A8-9600+1050. But it is an alternative.

Yea, I have been thinking the same thing. Seems like all the "quad cores are dead" talk has disappeared. How convenient. Hell, I even recall the 7700k being slammed as ready to fall off a cliff and similar comments. Now we are supposed to buy a quad core apu, and upgrade it later? Ok, whatever.

 

[denywinarto]

Personally I'll choose 2200G and add 1030/1050 in a year or two.

GPU is not an option..
I'd like to keep component count at minimum as possible due to electricity and depreciation cost reason..
AMD APU seems to be more than enough from my experience for the market here..

if you gonna use it in internet cafe then its better of with high frequency ram, because you can get good game performance now and then when you need more performance just sell the cpu and maybe by then there are zen2 apu based.

Is picasso APU confirmed or cancelled? cause i didn't see it on 2018 roadmap leaks

 

[frozentundra123456]

Perhaps 14++ and 14+ are different but 14 and 14+ are the exact same process. It is just marketing PR by intel. I don't even think 14++ is out yet so obviously that could be a real difference and I never said anything about 14++

So again I was not "absolutely wrong". Read my post. I said not a word about 14++ and I have seen ZERO documentation that says there is any difference between the process used to manufacture Skyalke(14) and the one used to manufacture Kaby Lake and Coffee Lake (14+)

Hate to tell you, but Coffee Lake IS 14nm++.

 

[FIVR]

Hate to tell you, but Coffee Lake IS 14nm++.

Wow, that's really disappointing then. I figured they could get much more improvement by actually improving the process and lowering the contact pitch, instead of just making up a new name for PR purposes like they did with 14nm+ and Kaby Lake. I guess that says a lot about why they can't get anything done on 10nm - their process engineers are awful.

 

[prtskg]

Yea, I have been thinking the same thing. Seems like all the "quad cores are dead" talk has disappeared. How convenient. Hell, I even recall the 7700k being slammed as ready to fall off a cliff and similar comments. Now we are supposed to buy a quad core apu, and upgrade it later? Ok, whatever.

That's enthusiasts talking, it's about very good performance. Here we are talking about budget builds, where performance needs to be good enough.
Enthusiasts prefer 6/8 cores. For budget builds, 2 or 4 cores with integrated gpu is norm.
Here AAA gaming is privilege where as in the group of quad cores are dead, that's a norm.
Though I'm sure, you already knew that.

 

[AtenRa]

OC+DDR4-3600? Someone is moving the goalposts here.

Thats not me, from the start i was talking about 1500-1600MHz and 3600 Memory.

On paper the 2200G has to share that bandwidth with the CPU. Also the 1030 is nothing, it can barely hold its own on 720p, just like the RX550.
If it matches it, fine, but it will still get crushed by a 1050, so it really dosent worth to pay any extra for it.
I still find hard to belive it can even match the 1030, AMD IGPs are yet to reach R7 250 GDDR5 levels of performance, GT1030/RX550 is a bit far off. And those slides does not give me any hope at all, why will they show the 2400G vs the I5-8400 IGP, and then show you it can match the I5-8400 + GT1030 on firestrike? There can be only one reason for that, and you know it.

Not to mention about 100% over the A12-9800 should be at 72 font size on a slide, they did not even mention Bristol Ridge.

You know very well what is the reason for AMD hidding stuff(or being very very creative with the fight they pick) on the slides in the past.

I will tell you this,

Ryzen 2200G ($100) OC 1500-1600MHz + 2x 4GB DDR-4 3600 ($120) = Total of $220

Now find me any alternative you want with the same CPU and GPU performance + 2x 4GB Memory for the same price.

ps. I will not even mention the ability to upgrade to a 8C 16T CPU in the future or the ability to build a USFF system.

 

[tamz_msc]

and I thought that consensus from previous discussions was 4/4 cores were dead for gaming, so why we are even considering a 4/4 2200G to be viable in any way now?

If anyone considers the G4560 as an alternative, the A8-9600 is also an alternative, its simple as that. And for gaming the A8-9600 paired with a 1050 will just destroy the 2200/2400G alone. The 9600 will have a way faster GPU avalible, more ram avalible (6 vs 8), and its cores are not going to be fighting the gpu for memory bandwidth, so even if they are slower cores, it worth it to have faster perf now, and you can upgrade later to a Ryzen CPU. Yes, a 2200G+1050 its going to be better than A8-9600+1050. But it is an alternative.

Yea, I have been thinking the same thing. Seems like all the "quad cores are dead" talk has disappeared. How convenient. Hell, I even recall the 7700k being slammed as ready to fall off a cliff and similar comments. Now we are supposed to buy a quad core apu, and upgrade it later? Ok, whatever.

Some very nice straw man arguments you two have got going on there. On this forum, to the best of my knowledge, there is only one poster who insists on how 4C/4T CPUs are 'dead' because of how 99% frame rates are affected in something like 64 player BF1 online matches, and that too with GTX 1070 class GPUs or even greater. For the vast majority(almost 3/4th of the Steam user-base), the potential shortcoming of a quad-core CPU is of no concern at all.

If we're going down that route - cherry picking games based on certain characteristics to show how CPUs of a particular type fall behind - then I'm within my rights to say how in the latest fad that is PUBG, merely going from a 1050Ti to a 1060 induces a CPU bottleneck on the older construction cores. Arguably it is much more relevant and shows that 4C/4T CPUs are still not 'dead'.

Even in another popular game like GTA V, a Ryzen 4C/4T would be 50% faster than any of the 28nm APU(the i3 7100 is about the same as the Ryzen 3 1200) . If anything, it proves that older construction core are indeed dead, and telling somebody to choose an APU based on those cores over a Ryzen 4C/4T is actually some of the worst possible advice somebody can give right now to a budget-minded consumer who intends to play current games.

So to summarize,

• G4560+GTX 1050/Ti is viable only if you have access to a good 2nd hand market for an upgrade to an i7 4C/8T (which for people outside the US/Canada, Aus/NZ and a select few European countries is not worth it)
• Coffee Lake starting with the i3 8100 but we're still waiting for the cheaper motherboards.
• AM4 with 2200G - just considering it as a CPU-only, it is a steal for 100$ considering the improvements it has over the similarly-priced Ryzen 3 1200. Cheap motherboards are available right now and paired with the GTX 1050 it is a better long-term investment for most of the people around the world, even if you save 25$ choosing the G4560 instead.
• All 28nm APUs are totally not worth it at this point.

 

[french toast]

Most non PC gamers have no problem with console graphics.

We are talking about budget graphics or intergrated graphics...are you expecting 4k hdr ultra @ 60fps ? .....for budget conscious buyers, PS4 gaming is pretty good.

We aren't talking about your personal interpretation, we are talking about actual product.

There's nothing called 14nm+, either in GF processes, nor in any current AMD documentation.

GF 14nm+ only existed on early AMD roadmaps, which were later updated to 12nm.

Ryzen Mobile at AMD:
https://www.amd.com/en/products/apu/amd-ryzen-7-2700u
"
CMOS
14nm
"

14nm, not 14nm+.

We both know AMDs marketing division is away with the fairies and learns as it goes along from Nvidia/intel...their marketing for their products have very little strategy and changes from time to time depending on what the landscape is, making tracking it difficult.

Here is a slide with AMD mentioning 3rd generation 14nm lpp
https://m.hexus.net/tech/news/graph...-500-graphics-cards-presentation-slides-leak/
We know 14nm lpp is second generation, (first for AMD or "14nm")...Rx 580 3rd gen/ AMD 2nd gen 14nm ...would = 14nm + in intel speak....then 12nm would be 14nm++ in intel speak (4th generation amd speak)....BUT AMD looked at Nvidia creative 12nm marketing and swapped over to that....that is amd for you, incoherent marketing.

Thats not me, from the start i was talking about 1500-1600MHz and 3600 Memory.



I will tell you this,

Ryzen 2200G ($100) OC 1500-1600MHz + 2x 4GB DDR-4 3600 ($120) = Total of $220

Now find me any alternative you want with the same CPU and GPU performance + 2x 4GB Memory for the same price.

ps. I will not even mention the ability to upgrade to a 8C 16T CPU in the future or the ability to build a USFF system.

Your nailing this one, good point.

Some very nice straw man arguments you two have got going on there. On this forum, to the best of my knowledge, there is only one poster who insists on how 4C/4T CPUs are 'dead' because of how 99% frame rates are affected in something like 64 player BF1 online matches, and that too with GTX 1070 class GPUs or even greater. For the vast majority(almost 3/4th of the Steam user-base), the potential shortcoming of a quad-core CPU is of no concern at all.

If we're going down that route - cherry picking games based on certain characteristics to show how CPUs of a particular type fall behind - then I'm within my rights to say how in the latest fad that is PUBG, merely going from a 1050Ti to a 1060 induces a CPU bottleneck on the older construction cores. Arguably it is much more relevant and shows that 4C/4T CPUs are still not 'dead'.

Even in another popular game like GTA V, a Ryzen 4C/4T would be 50% faster than any of the 28nm APU(the i3 7100 is about the same as the Ryzen 3 1200) . If anything, it proves that older construction core are indeed dead, and telling somebody to choose an APU based on those cores over a Ryzen 4C/4T is actually some of the worst possible advice somebody can give right now to a budget-minded consumer who intends to play current games.

So to summarize,

• G4560+GTX 1050/Ti is viable only if you have access to a good 2nd hand market for an upgrade to an i7 4C/8T (which for people outside the US/Canada, Aus/NZ and a select few European countries is not worth it)
• Coffee Lake starting with the i3 8100 but we're still waiting for the cheaper motherboards.
• AM4 with 2200G - just considering it as a CPU-only, it is a steal for 100$ considering the improvements it has over the similarly-priced Ryzen 3 1200. Cheap motherboards are available right now and paired with the GTX 1050 it is a better long-term investment for most of the people around the world, even if you save 25$ choosing the G4560 instead.
• All 28nm APUs are totally not worth it at this point.

Correct, strawmans abound, no one would recommend construction cores over 99$ 2200g, that's absurd.
Although g4560 and 8100 + dgpu is a solid budget option.
Either 2200g+dgpu or 2400g + 3600mhz ram is also a viable option, both amd options offer overclocking and better future upgrades.
CPUs on AMD side would likely outperform intel budget options for majority of tasks, expect that 2400g 4/8thread with 3600mhz ram.

Intel has some solid options, but AMD looks to win the day here, Bristol ridge desktop is not in the discussion.

 

[coercitiv]

Yea, I have been thinking the same thing. Seems like all the "quad cores are dead" talk has disappeared. How convenient. Hell, I even recall the 7700k being slammed as ready to fall off a cliff and similar comments. Now we are supposed to buy a quad core apu, and upgrade it later? Ok, whatever.

Are you trying to tell everybody in this thread you honestly can't tell the difference between a high-end and low-end gaming machine?

Let me help you with a simple pointer: the high end gaming machine is the one where the dGPU alone costs as much as the entire low-end gaming machine. On that system the quad core is no longer a good idea, not if you intend to buy new and keep for more than 2 years, with incremental GPU upgrades.

 

[PeterScott]

Thats not me, from the start i was talking about 1500-1600MHz and 3600 Memory.



I will tell you this,

Ryzen 2200G ($100) OC 1500-1600MHz + 2x 4GB DDR-4 3600 ($120) = Total of $220

Now find me any alternative you want with the same CPU and GPU performance + 2x 4GB Memory for the same price.

ps. I will not even mention the ability to upgrade to a 8C 16T CPU in the future or the ability to build a USFF system.

All this really means is that the APU beats the Intel IGP, which isn't exactly news. Before Ryzen APU, you could have said the same thing about BD based APUs.

That still doesn't make it something worth upgrading to.

If I was looking at minimal money to beat this, I would just buy a new GPU for my old computer.