Question Speculation: Ryzen 3000 series pricing

[bsp2020]

There has been a lot of post back and forth about what the pricing and core counts will be of Zen2 based Ryzen processors. So, I think it's time to ask people to vote for what they believe AMD's high end AM4 product stack will look like at what prices. I think there are mainly two different opinions about what the high end AM4 offering will be like. So, please, vote for your favorite.

Edit: The extra comment I put at the end of the last choice does not make sense. Please, ignore it.
"They are both too cheap. AMD will go after market share at all cost."

 

[B-Riz]

JFC I am tired of posting this, look at the minimums. 6 core i5 and 8 core i7 is a short term investment, deal with it.

 

[amrnuke]

It's not about the frame rates,it's about the number of cores/threads.
If you argue that no game scales beyond 6 cores then the results aren't all that exiting but there is scaling between the 6 core and 8 core zens and a 6/6 cpu beating a 8/16 is pretty telling and it's still matching it even when the 8/16 is overclocked to 4.2 on all cores.
Why would any casual PC user choose a 16t CPU that can't match a 6c one in the only demanding thing that person is ever going to do?

A fair point. Anandtech compared the 2700X, 2700, 2600X, and 2600 with the 9400's predecessor (basically) the 8400, with the 1600 in there for good measure. You can review on your own, the point being that when the GPU is not the limit, for gaming, the 8400 is clearly and consistently on top, at a lower price point compared to the 2700X, which is its closest comparison from AMD. For productivity, AMD ahead. At 4K gaming, they're even, at 1080p Intel best.

However, I'd keep in mind that for multiplayer gaming where you're having to handle more than just the game (eg streaming and using separate communication software), none of those benchmarks will account for that.

Someone mentioned already that 6C/6T is short-sighted. It also can lead to stuttering performance at times.

In any case, I think the higher threaded processors provide a far better future-proofing than the lower-threaded ones. But the reasons I believe that may not apply to everyone's needs.

 

[DrMrLordX]

good thing it supports 3DNow!

Sold! And to think I wanted Zen2. Now I can replace my uh, hmm, K6-III?

It's coming down to do 1440 high refresh

Welcome to the club. Though eventually you're going to need something faster than the 1080Ti, sorry to say. It's all about those . . .

look at the minimums.

Yes, the minimums. And that's why I wouldn't take the 9400F too seriously. It isn't on that benchmark list, but the 6c/6t 8600k is, and it loses the minimum fps benchmark to the lowly R7 2700. The 8400 fares even worse (predictably).

 

[TheELF]

Most games really don't use more than 6T, but that will inevitably change. And there is certainly software that will take all the threads you can throw at it. Plus, having those extra threads means I don't really have to worry about other things running should I decide to start up a game.

Because the benchmarks only show single-player and core-count can have a noticeable effect in multiplayer.

However, I'd keep in mind that for multiplayer gaming where you're having to handle more than just the game (eg streaming and using separate communication software), none of those benchmarks will account for that.

To tackle all of these with just one game look at BF V.
It easily uses 12 threads meaning that neither the 9400f or any 6/12 ryzen is able to handle background apps without affecting FPS,you will start loosing FPS with the smallest background app on either one.
The difference is that with the 9400 you will have a 25% buffer in more FPS (min) that you can eat into before you reach ryzen FPS
https://gamegpu.com/action-/-fps-/-tps/battlefield-v-test-gpu-cpu-2018

If you have a game that uses all threads in single player then you are not getting away with not loosing performance in multiplayer.
Both CPUs loose FPS when running multiplayer and they still both have all threads occupied and the 9400f will still have a nice FPS buffer to loose before hitting 6/12 ryzen FPS.
https://www.techspot.com/review/1754-battlefield-5-cpu-multiplayer-bench/

 

[amrnuke]

To tackle all of these with just one game look at BF V.
It easily uses 12 threads meaning that neither the 9400f or any 6/12 ryzen is able to handle background apps without affecting FPS,you will start loosing FPS with the smallest background app on either one.
The difference is that with the 9400 you will have a 25% buffer in more FPS (min) that you can eat into before you reach ryzen FPS
https://gamegpu.com/action-/-fps-/-tps/battlefield-v-test-gpu-cpu-2018

If you have a game that uses all threads in single player then you are not getting away with not loosing performance in multiplayer.
Both CPUs loose FPS when running multiplayer and they still both have all threads occupied and the 9400f will still have a nice FPS buffer to loose before hitting 6/12 ryzen FPS.
https://www.techspot.com/review/1754-battlefield-5-cpu-multiplayer-bench/

First paragraph:
The GameGPU link is saying something a lot different than what you're saying as best I can tell. It does not test the 9400F, so we'll use the 8600K, which has a faster base and boost, and the same L3 cache, and the same 6C/6T setup.
FPS avg: dead even on DX11 at 150fps
Min FPS: i5 has a 122 vs 113 benefit, <10%.
However, CPU utilization at 6 cores for 8600K is 78%, vs 50% for Ryzen.
So the Ryzen has a LOT more overhead / cap space for multitasking.

Second paragraph:
Techspot dose include the 9400F's predecessor, the 8400. It has worse (5%) min performance than 2700X and 5% better avg FPS than 2700X. But Techspot didn't look at core utilization to give any indication of overhead space for multitasking. If we use the information from GameGPU, which is that the Ryzen isn't being pushed much at all compared to the more-powerful 8600K, I can only assume that as you start tacking on streaming or chat, Ryzen would lose less FPS than a less powerful 9400F. But we don't know, because Techspot didn't test core utilization, I'm just extrapolating.

 

[Arkaign]

The test setups at Techspot for the 9400f / 2600x review are kind of strange.

$220 x470 Gaming X and $120 H115i Water cooling for the 2600x.

$280 Aorus Master and stock cooler for the 9400f.

$109 16GB 3400/CL15 for both.

However, the 9400f is clearly a budget CPU, so I don't think anyone is going to be pairing it with basically $300 mobos. The only reason to go hog wild with a z370/390 selection is for extreme OC with the unlocked CPUs.

The $110 Asus Z370 Prime can easily push the 9400f at higher numbers than they got for their config. Toss a $20 212 clone on it and use Intel XTU to lock @ 4.1x6. The 9400 also doesn't really need crazy ram speeds, your budget $69 AR 3200 kit @ stock XMP will still be better @ 6C all core 4.1 than looking for the more expensive ram.

By going 9400/AsusZ370/212/3200, you get $340, basically a full $200 cheaper than the Techspot build, and faster to boot.

On the 2600x side, to stay at that same $340, it's totally doable with Aorus x470 $109, stock cooler, same 3200 16GB kit. Literally almost half the price of their frankly silly $630 combo with mega Mobo and watercooling.

Between the budget versions of these builds, the 9400f would definitely gain performance, while the budget 2600x would lose very little at about half the cost.

Normally I don't think test configurations are all that relevant, but the choices made there for obstensibly budget choices seemed slightly misleading to me. As with more realistic components for each build, the Intel would be slightly faster, the Ryzen ever so slightly slower with stock cooling and 3200 mem.

Now some people I'm sure do go out and pair a bananas Mobo with a budget processor, but I've personally never seen it, not in 30+ years in the industry.

 

[Shivansps]

Well that depends on country, here a budget gaming build consist of a H310/A320, a GTX1050 3GB, 8GB DDR4-2400 single channel ram and generic as hell PSU. With CPU options going from 200GE to 2600 for AMD and G5400 to 9400F for Intel. With ocasional swap of GPU when cheap RX570 4GB are avalible.

 

[Mockingbird]

any last minute guesses?

 

[jpiniero]

any last minute guesses?

4 core (Zen+ APUs) - 119/149
6 core - 179/199
8 core - 279/329
12 core - 449

 

[Zucker2k]

4 core (Zen+ APUs) - 119/149
6 core - 179/199
8 core - 279/329
12 core - 449

Pretty much.

 

[JDG1980]

any last minute guesses?

8-core chip with max clocks designed to match 9900K performance at $329-$349. Power-optimized 8-core chips (replacement for 1700/2700 non-X) will be closer to $249.
12-core chips at $399 for the power-optimized version, and a flagship $499 price point for the frequency-optimized version.

 

[piesquared]

any last minute guesses?

If there is a 12 and 16 core a reasonable ask is:

599/799

for a theoretical 8 and 10 core:

450/ 550

and down from there. Particularly if the performance matches or beats the competition. They don't need to price them any lower, unless intel's reacts by cutting their prices.

 

[Zucker2k]

If there is a 12 and 16 core a reasonable ask is:

599/799

for a theoretical 8 and 10 core:

450/ 550

and down from there. Particularly if the performance matches or beats the competition. They don't need to price them any lower, unless intel's reacts by cutting their prices.

I think it'll be a mistake to price client chips that high. $500 is the threshold, imho. If they price those chips that high, Intel could just leave things be unless the 8 core Ryzen handily beats the 9900k at a significantly lower price.

 

[JDG1980]

If there is a 12 and 16 core a reasonable ask is:

599/799

for a theoretical 8 and 10 core:

450/ 550

and down from there. Particularly if the performance matches or beats the competition. They don't need to price them any lower, unless intel's reacts by cutting their prices.

The 2700X debuted with a MSRP of $329. I don't see them upping prices by $120 for the same core count. Maybe $349 for the top 8-core model if it matches/beats 9900K. A majority of buyers will still go with Intel if perf/$ is the same. AMD isn't going to blow out their new chips at bargain-basement prices, but they are going to offer a more compelling value proposition than Intel.

Most of the rumors seem to indicate that 16-core chips won't be coming at first; the launch configurations for Ryzen 3000 series will top out at 12-core. I think that they will target the $499 price point for their launch flagship. Maybe $599 if they feel especially optimistic about it.

My bet is that the 12-core configuration becomes the new "Ryzen 7", the 8-core drops to "Ryzen 5", and the 6-core to "Ryzen 3". The architecture clearly permits 16-core chips on AM4 (since there's room for 2 chiplets) so I expect to see that further down the road, probably as "Ryzen 9" slotting in above the existing chips. They may drop prices on their 12-core and below to accomodate it; I could see $599, maybe even $649 for the 16-core chip as a realistic possibility. They won't go much higher than that because the 16-core Threadripper 2950X is already available for $799 and an AM4 chip will have less connectivity and memory bandwidth.

 

[OTG]

If they do the whole lineup:
R3 6/6: $99, $129
R5 6/12: $169, $189
R5 8/16: $229, $249
R7 12/24: $329, $399
R9 16/32: $499, $599(?). Not much more because Threadripper.

With a full node shrink, plus Zen2 being intended to compete against a functional 10nm, going +50% in core count seems realistic. Yields are allegedly decent, and I still think that AMD's going to go for the throat on pricing.
15% IPC gain is nice and hefty if legit, speeds are the biggest question.
If the 12-core is under $400, one of them will probably find a home on my x370 taichi.
It may be dumb, but I really want something to push this overkill VRM more than my 1600 can.

 

[amd6502]

$110 3200g 3.6-4.0 4c/4t 8CU ryzen 3
$125 3300g 3.3-3.7 4c/8t 8CU ryzen 3
$160 3400g 3.7-4.2 4c/8t 11CU ryzen 5
TBA
$210 6c 3600 4.0-4.4 6c/12t 7nm ryzen 7
$250 6c 3700 4.2-4.7 6c/12t 7nm ryzen 7
$280 8c 3800 4.0-4.6 8c/16t 7nm ryzen 7
$340 8c 3800x 4.3-4.7 8c/16t 7nm ryzen 7
$400 12c 3910 4.0 -4.6 12c/24t 7nm ryzen 9
$500 12c 3920x 4.2-4.8 12c/24t 7nm ryzen 9
TBA

TBA = Aug-Nov launches and releases:
$140 8c 3385x 3.4-4.1 8c/8t 12nm ryzen 3
$150 6c 3565 3.5-4.0 6c/12t 12nm ryzen 5
$160 8c 3585 3.0-4.0 8c/16t 12nm ryzen 7
$145 4c 3500 3.8-4.1 4c/8t 7nm ryzen 5
$175 4c 3500x 4.0-4.4 4c/8t 7nm ryzen 5
$600 16c 3950 3.8-4.5 16c/32t 7nm ryzen 9
$750 16c 3960x 4.1-4.7 16c/32t 7nm ryzen 9

 

[DrMrLordX]

No way AMD goes over $599 on the 16c AM4 chip.

 

[amd6502]

https://twitter.com/x/status/1132660727993364480

A highly credible source hints there will be a low bin 6c:

3.6 base 4.2 boost.

(This probably means I need at least a 200mhz downward revision for my 7nm guesses. And it means 4ghz will probably be the highest base clock on the best binned 8c and 6c chiplets only.)

 

[Mockingbird]

No way AMD goes over $599 on the 16c AM4 chip.

AMD can charge whatever it wants ($699, $799, etc) since its faces no competition from Intel at this market segment. (Core i9-7960X is $1699)

There is no reason why there would be an arbitrary limit.

 

[DrMrLordX]

There is no reason why there would be an arbitrary limit.

So they go from 8c @ $399 to 12c @ $499 to 16C @ $799? We'll see. It is true that Intel can't counter on the desktop. Once more, their HEDT products are under attack by AM4.

I will say there is historic precedent for a $850 price point for the 16c part if it can match a 7960X as you indicate. They matched the 6900k ($1k) with a $500 CPU. So they can pit an $850 CPU against Intel's $1700 chip if they perform about as well, just to make Intel look stupid. Again.

 

[TheELF]

First paragraph:
The GameGPU link is saying something a lot different than what you're saying as best I can tell. It does not test the 9400F, so we'll use the 8600K, which has a faster base and boost, and the same L3 cache, and the same 6C/6T setup.
FPS avg: dead even on DX11 at 150fps
Min FPS: i5 has a 122 vs 113 benefit, <10%.
However, CPU utilization at 6 cores for 8600K is 78%, vs 50% for Ryzen.
So the Ryzen has a LOT more overhead / cap space for multitasking.

Second paragraph:
Techspot dose include the 9400F's predecessor, the 8400. It has worse (5%) min performance than 2700X and 5% better avg FPS than 2700X. But Techspot didn't look at core utilization to give any indication of overhead space for multitasking. If we use the information from GameGPU, which is that the Ryzen isn't being pushed much at all compared to the more-powerful 8600K, I can only assume that as you start tacking on streaming or chat, Ryzen would lose less FPS than a less powerful 9400F. But we don't know, because Techspot didn't test core utilization, I'm just extrapolating.

However, CPU utilization at 6 cores for 8600K is 78%, vs 50% for Ryzen.
So the Ryzen has a LOT more overhead / cap space for multitasking.

It doesn't work like that, if every thread has some load then running something else on the same thread will impact either one of the tasks.
Use affinity on the ryzen and confine the game on 50% of the threads(or to 50% usage on all threads) ,you will get a very different result,you can't just use that 50% idle as if it's idle.

 

[TheELF]

The test setups at Techspot for the 9400f / 2600x review are kind of strange.

$220 x470 Gaming X and $120 H115i Water cooling for the 2600x.

$280 Aorus Master and stock cooler for the 9400f.

$109 16GB 3400/CL15 for both.

However, the 9400f is clearly a budget CPU, so I don't think anyone is going to be pairing it with basically $300 mobos. The only reason to go hog wild with a z370/390 selection is for extreme OC with the unlocked CPUs.

The $110 Asus Z370 Prime can easily push the 9400f at higher numbers than they got for their config. Toss a $20 212 clone on it and use Intel XTU to lock @ 4.1x6. The 9400 also doesn't really need crazy ram speeds, your budget $69 AR 3200 kit @ stock XMP will still be better @ 6C all core 4.1 than looking for the more expensive ram.

By going 9400/AsusZ370/212/3200, you get $340, basically a full $200 cheaper than the Techspot build, and faster to boot.

On the 2600x side, to stay at that same $340, it's totally doable with Aorus x470 $109, stock cooler, same 3200 16GB kit. Literally almost half the price of their frankly silly $630 combo with mega Mobo and watercooling.

Between the budget versions of these builds, the 9400f would definitely gain performance, while the budget 2600x would lose very little at about half the cost.

Normally I don't think test configurations are all that relevant, but the choices made there for obstensibly budget choices seemed slightly misleading to me. As with more realistic components for each build, the Intel would be slightly faster, the Ryzen ever so slightly slower with stock cooling and 3200 mem.

Now some people I'm sure do go out and pair a bananas Mobo with a budget processor, but I've personally never seen it, not in 30+ years in the industry.

They are testing dozens of CPUs in one go,they can't build the best system for each CPU,they go for the extreme build that will not hinder the best CPUs they have and then just switch out CPUs.

 

[amrnuke]

It doesn't work like that, if every thread has some load then running something else on the same thread will impact either one of the tasks.
Use affinity on the ryzen and confine the game on 50% of the threads(or to 50% usage on all threads) ,you will get a very different result,you can't just use that 50% idle as if it's idle.

Surely there will be an impact when using other apps, I agree. I'm not sure why you would assume that I'd think you can just pack whatever you want into that unused "load" and it'll just work. The point is that the 2600x would (theoretically) see less of a performance hit because it's not even close to full utilization, whereas due to the processor dependency of streaming, Twitch would probably eat into Battlefield's performance on 9400F quite a bit more than it would on the 2600x. However, one would have to have a very finely controlled test environment to prove that. I couldn't find where it had ever been done.

 

[TheELF]

Surely there will be an impact when using other apps, I agree. I'm not sure why you would assume that I'd think you can just pack whatever you want into that unused "load" and it'll just work. The point is that the 2600x would (theoretically) see less of a performance hit because it's not even close to full utilization, whereas due to the processor dependency of streaming, Twitch would probably eat into Battlefield's performance on 9400F quite a bit more than it would on the 2600x. However, one would have to have a very finely controlled test environment to prove that. I couldn't find where it had ever been done.

If each core already runs two threads then it's at full utilization.
You can look at the 4/8 zen part to see how much performance you loose by not having those threads running for your game.

Twitch would probably eat into Battlefield's performance on 9400F quite a bit more than it would on the 2600x.

Yes it would, that's what I said,but the 9400f will also have a nice FPS advantage to eat into.
The 8400 is 100Mhz slower then the 9400f and has a 135FPS avg compared to the 109FPS avg of the 2600x ,6 cores vs 6 cores with SMT and the 8400 still has a 24% advantage in avg FPS you can eat into.

 

[DrMrLordX]

It's all academic now. Nobody should choose a 2600x or a 9400F over an R5 3600x.