Post CFL Ryzen Price drop / SKU shuffle thoughts, predictions?

majord

Senior member
Jul 26, 2015
433
523
136
So I haven't seen much discussion on this, but now we know Coffeelake lineup and pricing, what's Everyone's prediction,/ thoughts on AMD's inevitable pricing and SKU shuffle in response?

Personally I think some SKU's need to go, and everything shoved down a tier, until the refresh comes in 2018. Before shooting down my suggestions too early, bear in mind I've based my own thoughts on the fact Ryzen does not inculde any IGP, and a continued strategy of [slightly] undercutting intel at all times for comparable performance.

Ryzen 3 line: Completely replace with SMT enabled SKU's. I don't see any way these can remain as 4c/4t without dropping them down into sub $100 mark.. I guess that's also a possibilty.

Ryzen 5 line: 6c/12c SKU's only. prices dropped down a tier, small clock bumps.

Ryzen 7 line: 1700 - retain as 65w but with a small clock bump if yields permit
1700x - discontinue. 3's a crowd.
1800x : pretty much has to stay has top tier, but with price dropped to $349.

anyone else?
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
So I haven't seen much discussion on this, but now we know Coffeelake lineup and pricing, what's Everyone's prediction,/ thoughts on AMD's inevitable pricing and SKU shuffle in response?

Personally I think some SKU's need to go, and everything shoved down a tier, until the refresh comes in 2018. Before shooting down my suggestions too early, bear in mind I've based my own thoughts on the fact Ryzen does not inculde any IGP, and a continued strategy of [slightly] undercutting intel at all times for comparable performance.

Ryzen 3 line: Completely replace with SMT enabled SKU's. I don't see any way these can remain as 4c/4t without dropping them down into sub $100 mark.. I guess that's also a possibilty.

Ryzen 5 line: 6c/12c SKU's only. prices dropped down a tier, small clock bumps.

Ryzen 7 line: 1700 - retain as 65w but with a small clock bump if yields permit
1700x - discontinue. 3's a crowd.
1800x : pretty much has to stay has top tier, but with price dropped to $349.

anyone else?
I totally agree, but the 1800x price needs to drop to $325 in the mainstream segment; it's strengths are niche in this segment. I agree that the 1700x needs to go.
If I were AMD, I'll keep only one octocore in the mainstream segment (it'll be the ONLY octocore in this segment, for now); the R7 1700 or 1800x? Preferably, the 1800x. Prices on Newegg are $299.99 and $449.99 respectively. Whichever chip they decide on, they should price it at $300 and call it a day. Or, price the 1800x at $325, move the 1700 to $265, and move the 1600x to $199.99.

Edited for clarity.
 

epsilon84

Golden Member
Aug 29, 2010
1,142
927
136
What you are suggesting is a major overhaul of the Ryzen lineup in response to CFL. I don't think AMD will do anything so drastic. Price cuts are possible but I don't imagine there will be a reshuffling of SKUs.

Ryzen 3 with 4C/8T is essentially a Ryzen 5 1400/1500, so that won't happen unless they scrap Ryzen 3 altogether.

Will be interesting to see how R7 chips are priced, I think approx $300 USD for the 1800X would be fair considering the price of the 8700 (non K), you're looking at probably a slight MT advantage to the 1800X but a significant ST deficit plus, as you said, no IGP.

Also interested to see how R5 pricing will be affected, the sub $200 i5 8400 should (in theory) drive down the prices of the 1600/1600X, making them even better value than they are today.
 

Rifter

Lifer
Oct 9, 1999
11,522
751
126
What you are suggesting is a major overhaul of the Ryzen lineup in response to CFL. I don't think AMD will do anything so drastic. Price cuts are possible but I don't imagine there will be a reshuffling of SKUs.

Ryzen 3 with 4C/8T is essentially a Ryzen 5 1400/1500, so that won't happen unless they scrap Ryzen 3 altogether.

Will be interesting to see how R7 chips are priced, I think approx $300 USD for the 1800X would be fair considering the price of the 8700 (non K), you're looking at probably a slight MT advantage to the 1800X but a significant ST deficit plus, as you said, no IGP.

Also interested to see how R5 pricing will be affected, the sub $200 i5 8400 should (in theory) drive down the prices of the 1600/1600X, making them even better value than they are today.

I agree, we are only 6 months out from PR so i dont see AMD doing much with current ryzen lineup when its about to be replaced in 6 months anyways. Some price drops for sure, but no shuffling of SKU's.
 

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
Until Pinnacle Ridge arrives AMD might have to make some price cuts. With Pinnacle Ridge AMD can bump up prices back a bit but I would still like AMD to keep the Ryzen stack prices from USD 99 to USD 399. The Ryzen 7 stack should ideally be priced at USD 399/USD 349/USD 299 with Ryzen 7 2800x/2700x/2700.

R7 1800x - USD 349
R7 1700x - USD 299
R7 1700 - USD 249
R5 1600x - USD 209
R5 1600 - USD 179
R5 1500x - USD 139
R5 1400 - USD 119
R3 1300x - USD 99
R3 1200 - USD 79
 

R0H1T

Platinum Member
Jan 12, 2013
2,582
162
106
Until Pinnacle Ridge arrives AMD might have to make some price cuts. With Pinnacle Ridge AMD can bump up prices back a bit but I would still like AMD to keep the Ryzen stack prices from USD 99 to USD 399. The Ryzen 7 stack should ideally be priced at USD 399/USD 349/USD 299 with Ryzen 7 2800x/2700x/2700.

R7 1800x - USD 349
R7 1700x - USD 299
R7 1700 - USD 249
R5 1600x - USD 209
R5 1600 - USD 179
R5 1500x - USD 139
R5 1400 - USD 119
R3 1300x - USD 99
R3 1200 - USD 79
Is there a need for regular 4c4t R3 with RR also coming in next year?
 

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
Is there a need for regular 4c4t R3 with RR also coming in next year?
Yeah I expect AMD to discontinue the quads once Raven Ridge arrives.
I don't see how it matters really. Both Summit Ridge & RR have die around the same size, ~200mm^2, with the advantage going to SR. Seems it costs AMD the same either way to produce a 4 core CPU.
 

R0H1T

Platinum Member
Jan 12, 2013
2,582
162
106
I don't see how it matters really. Both Summit Ridge & RR have die around the same size, ~200mm^2, with the advantage going to SR. Seems it costs AMD the same either way to produce a 4 core CPU.
After RR they could simply sell the defective SR as athlon x2/4 or something, SR cannot occupy the same space with RR as there's no IF penalty for the latter & might (over)clock higher since it's not MCM.
 

Asterox

Golden Member
May 15, 2012
1,026
1,775
136
Yeah I expect AMD to discontinue the quads once Raven Ridge arrives.

Your have wrong expectations, AMD 4/4 CPU not going anywhere or in the grave etc.AMD Pinnacle Ridge 4/4, it will be even beter or improved new 4/4 CPU model.You can expect higher stock CPU frequency+higher overclocking potential or 4.3ghz.

There is no any logic to kill AMD-s 2017/100$ improved version of Intel i5 2500K(new 200$) from 2011.

Raven Ridge APU will not be dirty cheep, for the cheapest 4/4 Quad Core you can expect minimal price 150$.:cool:
 

VirtualLarry

No Lifer
Aug 25, 2001
56,226
9,990
126
R7 1800x - USD 349
R7 1700x - USD 299
R7 1700 - USD 249
R5 1600x - USD 209
R5 1600 - USD 179
R5 1500x - USD 139
R5 1400 - USD 119
R3 1300x - USD 99
R3 1200 - USD 79
Not sure that I agree with all of these. I could see 1300X going down to $110, and 1200 going down to $90. Also, 1600 dropping down to $200, and 1600x dropping to $215.

Raven Ridge APU will not be dirty cheep, for the cheapest 4/4 Quad Core you can expect minimal price 150$.:cool:
Yeah, RR won't be as cheap as Ryzen 3 non-SMT quads, that's for sure. I expect that Bristol Ridge will actually live on, just to eat up fab capacity on that process, and to bolster the bottom-line prices of the RR APUs.
 

majord

Senior member
Jul 26, 2015
433
523
136
What you are suggesting is a major overhaul of the Ryzen lineup in response to CFL. I don't think AMD will do anything so drastic. Price cuts are possible but I don't imagine there will be a reshuffling of SKUs.

Ryzen 3 with 4C/8T is essentially a Ryzen 5 1400/1500, so that won't happen unless they scrap Ryzen 3 altogether.

Will be interesting to see how R7 chips are priced, I think approx $300 USD for the 1800X would be fair considering the price of the 8700 (non K), you're looking at probably a slight MT advantage to the 1800X but a significant ST deficit plus, as you said, no IGP.

Also interested to see how R5 pricing will be affected, the sub $200 i5 8400 should (in theory) drive down the prices of the 1600/1600X, making them even better value than they are today.

I realize it's a bit extreme, but it comes down to whether AMD want to retain that value edge, to even half the degree currently offered..

Mulling over 1300x vs i5 7500 benchmarks as a bit of reference point for R3 vs the new i3, it (ryzen) is not much slower, but it's still undoubtedly slower. overclocked it would probably only match or slightly nudge ahead of an 8100.. it does not have an IGP though. Hence why I think these SKU's if not discontonued would be destined for the sub $100 bracket

This brings up a whole other area of discussion though - what value can you put on an IGP that's become so incapable?.. The fact Ryzen has no igp is of little consequence to anyone wanting even 'low end' gaming performance at this point in time, since current Intel IGP's on desktop chips have not improved since skylake, pushing them further into the category of 'useless' for gaming. This means a discrete GPU will be used anyway, and the IGP situation will perceived value of the chip.

On the other end of the spectrum, Having no IGP also completely invalidates Ryzen as a choice for budget non gaming systems. So in this case, discounting it because of no IGP still won't sell chips to this market - no IGP = no sale.

I don't see how it matters really. Both Summit Ridge & RR have die around the same size, ~200mm^2, with the advantage going to SR. Seems it costs AMD the same either way to produce a 4 core CPU.

As per above.. the difference is one will have an IGP, one will have nothing but a dead silicon. Having an IGP - any IGP means these can be sold into the non-gaming , home/office PC market - having a capable IGP like RR means this market extends to casual / e-sports gamers as well.
 

Topweasel

Diamond Member
Oct 19, 2000
5,436
1,654
136
I realize it's a bit extreme, but it comes down to whether AMD want to retain that value edge, to even half the degree currently offered..

Mulling over 1300x vs i5 7500 benchmarks as a bit of reference point for R3 vs the new i3, it (ryzen) is not much slower, but it's still undoubtedly slower. overclocked it would probably only match or slightly nudge ahead of an 8100.. it does not have an IGP though. Hence why I think these SKU's if not discontonued would be destined for the sub $100 bracket

This brings up a whole other area of discussion though - what value can you put on an IGP that's become so incapable?.. The fact Ryzen has no igp is of little consequence to anyone wanting even 'low end' gaming performance at this point in time, since current Intel IGP's on desktop chips have not improved since skylake, pushing them further into the category of 'useless' for gaming. This means a discrete GPU will be used anyway, and the IGP situation will perceived value of the chip.

On the other end of the spectrum, Having no IGP also completely invalidates Ryzen as a choice for budget non gaming systems. So in this case, discounting it because of no IGP still won't sell chips to this market - no IGP = no sale.



As per above.. the difference is one will have an IGP, one will have nothing but a dead silicon. Having an IGP - any IGP means these can be sold into the non-gaming , home/office PC market - having a capable IGP like RR means this market extends to casual / e-sports gamers as well.

There are other reasons. Faster clocks. Less power usage. More cache. More PCIe.

People think too simplistic. This chip has a use. This one doesn't. They all have uses. 2 big reasons to still offer the R3's. 1. If they are selling AMD will continue to sell it. 2. It is a configuration of the current Zeppelin die that they can sell. These dies don't compete with use in the 1900x and the EPYC 7281 is the only EPYC that potentially could use these dies. The 7281 is probably the least interesting chip in the whole EPYC lineup, with little cost savings over the other 16c configurations that have 64MB cache.
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,686
136
This brings up a whole other area of discussion though - what value can you put on an IGP that's become so incapable?.. The fact Ryzen has no igp is of little consequence to anyone wanting even 'low end' gaming performance at this point in time, since current Intel IGP's on desktop chips have not improved since skylake, pushing them further into the category of 'useless' for gaming. This means a discrete GPU will be used anyway, and the IGP situation will perceived value of the chip.
For anything gaming related the iGPU is irrelevant: if you're going to run those games and want a (minimal) decent experience there are quite a few options in the budget dGPU area, starting with the low end GT 1030 / RX 550 that offer over 2X performance of HD 530.

On the other end of the spectrum, Having no IGP also completely invalidates Ryzen as a choice for budget non gaming systems. So in this case, discounting it because of no IGP still won't sell chips to this market - no IGP = no sale.
This is where Zen based APUs need to slot in ASAP, cutdown SKUs will finally address the non gaming budget market.

However, I wouldn't worry much about the R3 line, these SKUs are in short supply by design. If R3 supply start building up because Coffe Lake / Kaby Lake (Pentium) are getting the sales instead, all AMD needs to clear their inventory is to gradually decrease prices until they start moving again. It's probably the least of their worries.

The only parts that really matter are the R5 1600 and the R7 1700. These are the big sellers, they need to be adjusted to stay competitive, but AMD may just wait a bit to see how Intel's supply holds up. If it doesn't... they get to keep their prices for a while longer.
 
  • Like
Reactions: VirtualLarry

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
.............................................................
As per above.. the difference is one will have an IGP, one will have nothing but a dead silicon. Having an IGP - any IGP means these can be sold into the non-gaming , home/office PC market - having a capable IGP like RR means this market extends to casual / e-sports gamers as well.
The discussion was about the R3 line, meaning quad cores without IGP and if AMD would use RR or SR as the source of these CPUs. Bringing up RR with a working IGP is irrelevant to this product, an IGP-less quad.
 

R0H1T

Platinum Member
Jan 12, 2013
2,582
162
106
The discussion was about the R3 line, meaning quad cores without IGP and if AMD would use RR or SR as the source of these CPUs. Bringing up RR with a working IGP is irrelevant to this product, an IGP-less quad.
So what happens to RR with a defective IGP? Let's assume there could be some with fully non funtional IGP, wouldn't a 4c8t Zen that could possibly clock higher & not have the IF penalty, be worth so much more than the current R3?

I'm not necessarily arguing the branding but two defective 2c2t Zen glued together are a lot less appealing than a true quad core, 4c8t or 4c4t, & I assume there'd be many once the APU are released.
 
Last edited:

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
So what happens to RR with a defective IGP? Let's assume there could be some with fully non funtional IGP, wouldn't a 4c8t Zen that could possibly clock higher & not have the IF penalty, be worth so much more than the current R3?

I'm not necessarily arguing the branding but two defective 2c2t Zen glued together are a lot less appealing than a true quad core, 4c8t or 4c4t, & I assume there'd be many once the APU are released.
I'm fairly certain that the 2C2T CCX is an AMD marketing restriction. The % area needed for simultaneous multithreading is very small, so defective die numbers will be tiny. A 4C4T product can come from either SR or RR wafers. I actually see them staying with SR because:

[1] Both RR & SR cost about the same to fab with SR being marginally cheaper.
[2] Harvested RR will have 2C4T + IGP. RR is said to have "up to 4 cores" + various CU numbers, so there is a fair sized need for non fully functional die.
[3] A way to maximize yields on both products.
 

R0H1T

Platinum Member
Jan 12, 2013
2,582
162
106
I'm fairly certain that the 2C2T CCX is an AMD marketing restriction. The % area needed for simultaneous multithreading is very small, so defective die numbers will be tiny. A 4C4T product can come from either SR or RR wafers. I actually see them staying with SR because:

[1] Both RR & SR cost about the same to fab with SR being marginally cheaper.
[2] Harvested RR will have 2C4T + IGP. RR is said to have "up to 4 cores" + various CU numbers, so there is a fair sized need for non fully functional die.
[3] A way to maximize yields on both products.
That's one of the assumptions I made, but am not 100% certain. Then there's also the branding problem, where would 2c2t, 2c4t or 4c8t (RR) fit with the current R3 lineup.

The harvested RR, 4c4t, would likely be better than current R3 due to being true quad cores & no IF hassles. All I am saying is that I don't see the long term viability of 2 CCX R3, given ~
  1. the native quad core RR (4c4t or 4c8t) would almost certainly be better & quite likely OC better as well.
  2. the R3 realistically occupies the sub 100$ range, at that price IGP parts sell better IMO.
The single CCX dies (2c2t or 2c4t) should ideally be relegated to the class of Athlons or Semprons, after RR is released. That leaves a really tiny segment which the R3 could serve, give quad core i3 CFL is upon us.
 

Abwx

Lifer
Apr 2, 2011
10,847
3,297
136
Big difference between SR and RR is that the latter will use a shrinked version of Carrizo/Bristol Ridge integrated PCH, wich will imply much lower idle power than SR, so it should be preferably used even for GPU less 4C/4T CPUs.

Besides i m not sure if SR based 4C/4T are not fully functional 4C/8T dies with SMT disabled, as a mean to have a 100-120$ offering to fill the i3/i5 segment in the waiting of RR.
 

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
Big difference between SR and RR is that the latter will use a shrinked version of Carrizo/Bristol Ridge integrated PCH, wich will imply much lower idle power than SR, so it should be preferably used even for GPU less 4C/4T CPUs.

Besides i m not sure if SR based 4C/4T are not fully functional 4C/8T dies with SMT disabled, as a mean to have a 100-120$ offering to fill the i3/i5 segment in the waiting of RR.
Assumptions are that AMD wants to extract as much value as possible and production costs are similar for both.
SR die:
4C allows deeper harvesting. Stopping at 6 cores would discard some partially working die.

RR die:
A 4C non-IGP model would almost certainly have some of the GPU section working, meaning rejecting value if sold without IGP.
There will be 2C RR models to allow core harvesting
Ditto for various CU IGP models for GPU harvesting

Bristol Ridge and earlier models FM? model CPUs were an anomaly, as AMD had no other way of making a FM? motherboard compatible 4C non-IGP CPU except from the APU seeing that AM3+ was the other incompatible CPU line. This time AM4 is used by both pure CPUs and APUs.

In the end, only my opinion but it seems the most profitable solution to continuing a 4C4T & 4C8T non-IGP line.
 

jpiniero

Lifer
Oct 1, 2010
14,510
5,159
136
I don't see how it matters really. Both Summit Ridge & RR have die around the same size, ~200mm^2, with the advantage going to SR. Seems it costs AMD the same either way to produce a 4 core CPU.

It's more about removing a redundancy than anything else. Especially since AMD seems really reluctant to go under a hundred bucks it seems like replacing the quads with RR will help justify keeping the prices up.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
AMD's position will be chiefly determined by their market experiences from the past ten years or so. Also the WSA. I'm not sure what they intend to do with all the 14nm LPP wafers GF will continue producing, unless they can convince GF to go easy on them and make a primary switch to 12nm LP via retooling of fabs currently producing 14nm LPP.

WSA aside, expect AMD to lower prices to clear out the channels so that they aren't stuffed full of 2017's Summit Ridge when Pinnacle Ridge hits the scene in 2018. Overall, AMD's position is pretty strong since Intel hasn't even been able to get a cogent response to Ryzen to market until October, with a lot of the better Z370 boards lagging until November. We won't even see a proper platform upgrade in the form of Z390 until H2 2018 depending on whom you believe. Though I would expect closer to July than December, but you never know.

AMD had half a year to sell Ryzen against Kabylake, and depending on when 12nm LP is ready for non-risk production, Intel may have only 6-9 months to pit 6c Coffeelake against Ryzen. Overall AMD is in pretty good shape here.

I think 8c/16t Pinnacle Ridge - which is the best we'll probably see on AM4 - vs 8c/16t Coffeelake should be a very interesting contest.
 

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
It's more about removing a redundancy than anything else. Especially since AMD seems really reluctant to go under a hundred bucks it seems like replacing the quads with RR will help justify keeping the prices up.
Kinda confused here. Are you saying that RR will perform better than SR thus justifying higher prices? In any case, some quads will definitely be under $100. There is a large growing market outside the developed world.