Speculation: Ryzen 3000 series

[somethingclever]

Exactly.

I expect the 8C/16T Ryzen 7 3700X to cost around the same as the 8C/16T Ryzen 7 2700X.

Let me remind you.

What's incorrect though is making the assumption that it will be 8C/16T

 

[Mockingbird]

Let me remind you.

...and why shouldn't I assume that?

12C/24T and 16C/32T "Ryzen 9" would be a reasonable assumption, also

 

[maddie]

In 2006 Intel introduced the Core 2 Duo E6300 model. One AMD model at the time was the Athlon 64 X2 5000+. These two were pretty much equivalent in general usage (255 vs 259 Sysmark 2004) and gaming.

Prices were $183 Core 2 Duo E6300 and $ 403 Athlon 64 X2 5000+. This was Intel recovering from the disastrous Pentium 4 era.

Anandtech:

Final Words

Intel's Core 2 Extreme X6800 didn't lose a single benchmark in our comparison; not a single one. In many cases, the $183 Core 2 Duo E6300 actually outperformed Intel's previous champ: the Pentium Extreme Edition 965. In one day, Intel has made its entire Pentium D lineup of processors obsolete. Intel's Core 2 processors offer the sort of next-generation micro-architecture performance leap that we honestly haven't seen from Intel since the introduction of the P6.


Compared to AMD's Athlon 64 X2 the situation gets a lot more competitive, but AMD still doesn't stand a chance. The Core 2 Extreme X6800, Core 2 Duo E6700 and E6600 were pretty consistently in the top 3 or 4 spots in each benchmark, with the E6600 offering better performance than AMD's FX-62 flagship in the vast majority of benchmarks. Another way of looking at it is that Intel's Core 2 Duo E6600 is effectively a $316 FX-62, which doesn't sound bad at all.


note "In many cases, the $183 Core 2 Duo E6300 actually outperformed Intel's previous champ". The slowest (E6300) new model.

If Intel had followed the present advice of a few here, would recent history have been similar.

 

[jpiniero]

The difference is that all jacking up core counts will do is get people to buy the cheapest model. It's not like servers where those customers could use and would value extra cores.

 

[somethingclever]

12C/24T and 16C/32T "Ryzen 9" would be a reasonable assumption, also

It doesn't make sense for both 12C and 16C to be R9

 

[mattiasnyc]

Why does it seem like bad business to have found a way to produce products more efficiently?

That's not what I proposed though, is it?

revenue - cost of production and other expenses = profit

You're a business. The above is true. You lower production costs. Do you lower revenue by lower price and maintain the same profit, or do you maximize profit by NOT lowering price? Between the two is the calculation of whether or not increased sales from a lower price yields more profit. "We" are saying prices won't be that low.

Nowhere have I questioned what you imply.

 

[Mockingbird]

It doesn't make sense for both 12C and 16C to be R9

Why not?

Core i9 has anything from 8C/16T to 18C/36T.

 

[mattiasnyc]

If Intel had followed the present advice of a few here, would recent history have been similar.

The difference is found in your second sentence in that post:

This was Intel recovering from the disastrous Pentium 4 era.

AMD is not in that situation currently.

 

[Mockingbird]

AMD is not in that situation currently.

Exactly. That would be when AMD first launched Ryzen back in 2017.

 

[PeterScott]

Why does it seem like bad business to have found a way to produce products more efficiently? The industry has been researching using a chiplet strategy to address increasing node costs, and it seems that AMD beat everyone (as far as I know) to the punch. Even better, they used it to increase yields too.

He didn't say finding a way to produce more efficiently is bad business.

He said bargain basement pricing is bad business.

 

[OTG]

IMO, because then their top R7

The difference is that all jacking up core counts will do is get people to buy the cheapest model. It's not like servers where those customers could use and would value extra cores.

A rational consumer, maybe, but how many people here have a PC that's more than they "need"?
I sure do. There's no need at all for me to upgrade, but I'm still planning to.

I don't think they'll go past a 12-core for launch though.
8 cores seems to be the sweet spot right now between high clock speeds and multithreading.
12 cores is a nice +50%, 16 cores is probably worth holding back and releasing an R9, or a special edition.
Including 6-cores for the R3 line means they get to bump core counts across the board, which is a nice marketing blurb, and the chiplets means they can make a 6-core out of almost anything. Even two chiplets that each have 5 defective cores out of 8 could be used.
What seems like a waste of money (increasing core count) may cost almost nothing.

 

[PeterScott]

In 2006 Intel introduced the Core 2 Duo E6300 model. One AMD model at the time was the Athlon 64 X2 5000+. These two were pretty much equivalent in general usage (255 vs 259 Sysmark 2004) and gaming.

Prices were $183 Core 2 Duo E6300 and $ 403 Athlon 64 X2 5000+. This was Intel recovering from the disastrous Pentium 4 era.

If Intel had followed the present advice of a few here, would recent history have been similar.

Not at all. Intel increased per core performance and, maintained its same per core pricing structure. Intel did not lower per/core prices.

This is essentially exactly what I am saying AMD will do. They will improve per/core performance, but will not lower per/core pricing.

While others think AMD will increase per core performance, while slashing per core pricing at the same time. They will be disappointed, proven wrong IMO.

 

[OTG]

Not at all. Intel increased per core performance and, maintained its same per core pricing structure. Intel did not lower per/core prices.

This is essentially exactly what I am saying AMD will do. They will improve per/core performance, but will not lower per/core pricing.

While others think AMD will increase per core performance, while slashing per core pricing at the same time. They will be disappointed, proven wrong IMO.

Stop focusing so much on per-core pricing; what matters is that Intel radically improved performance, and sold it for less than half the price of AMD.
If Intel thinks it's worthwhile to undercut a competitor, surely AMD can do the same, and AMD has a much worse starting position than Intel did in 2006.
You think they will do it by matching core counts for less, I think their costs are low enough that they can do the same performance (8-core) for less, better performance (12-core) for about the price where the 9900k will end up, then crazy performance for about the 9900k's current price(theoretical R9 16-core).
They need this gain much, MUCH, more than Intel did when AMD briefly had the performance crown.

 

[maddie]

The difference is found in your second sentence in that post:



AMD is not in that situation currently.

Yes they are.

They have slightly more marketshare and sales than a couple years ago, and have only had a few profitable quarters, when all they had was Bulldozer era designs. Intel was profitable throughout their period of Pentium sales and also still dominated in marketshare.

 

[exquisitechar]

IMO, because then their top R7


A rational consumer, maybe, but how many people here have a PC that's more than they "need"?
I sure do. There's no need at all for me to upgrade, but I'm still planning to.

I don't think they'll go past a 12-core for launch though.
8 cores seems to be the sweet spot right now between high clock speeds and multithreading.
12 cores is a nice +50%, 16 cores is probably worth holding back and releasing an R9, or a special edition.
Including 6-cores for the R3 line means they get to bump core counts across the board, which is a nice marketing blurb, and the chiplets means they can make a 6-core out of almost anything. Even two chiplets that each have 5 defective cores out of 8 could be used.
What seems like a waste of money (increasing core count) may cost almost nothing.

Hmm, I think it would be a mistake to delay the release of the 12 core CPU if they don't have to for whatever reason. I don't expect the 8 core to be significantly faster than the 9900K in most workloads, and it will likely be slower in some, if not many...we cannot be sure of anything with just that Cinebench comparison. Of course, it will have much better price/performance, but a higher core count SKU in the 9900K's price range that is the undisputed performance champion in the majority of cases would be worth releasing ASAP. Clocks don't have to be much low(er) than the 8 core, it would just have a lower base clock. The turbo would probably be higher, as we've seen with Threadripper.

 

[mattiasnyc]

Yes they are.

I don't think AMD is still 'recovering'. Look at all the tech companies there are out there and pay attention to stock price increases for 2018.

 

[maddie]

He didn't say finding a way to produce more efficiently is bad business.

He said bargain basement pricing is bad business.

It must be a sort of Stockholm syndrome effect in play. Intel hijacks your wallet and we sympathize with their pricing.

Maybe, just maybe, AMD prices won't be basement ones. They just won't be sky-high ones. Using the present Intel as the lords of reasonable pricing is astounding.

 

[exquisitechar]

I don't think AMD is still 'recovering'. Look at all the tech companies there are out there and pay attention to stock price increases for 2018.

Who cares about the stock price? Look at the market share and revenue, AMD still has a long way to go.

 

[maddie]

I don't think AMD is still 'recovering'. Look at all the tech companies there are out there and pay attention to stock price increases for 2018.

That price is based on future expectations for the company, aka when they really have recovered. AMD is one of the few growth stocks in an established market. That is why they have to act like a startup instead of solely like an established one. Those who don't want them to succeed will argue strongly for them to go slow, no need to disrupt, etc, all the while Intel is scrambling back.

 

[OTG]

I don't expect the 8 core to be significantly faster than the 9900K in most workloads, and it will likely be slower in some, if not many...we cannot be sure of anything with just that Cinebench comparison.

Isn't that exactly why they DO need to release the 12-core?
Matching the 9900k is nice, but once Intel drops prices, it's 50/50 which you choose, and that favors Intel.
If the R7 is 12-cores and performs +30-50% above the 9900k, boom, AMD wins, and it's not even close.

Of course, it will have much better price/performance, but a higher core count SKU in the 9900K's price range that is the undisputed performance champion in the majority of cases would be worth releasing. Clocks don't have to be much low(er) than the 8 core, it would just have a lower base clock. The turbo would probably be higher, as we've seen with Threadripper.

The 9900k won't stay over $500, so the pertinent number is where the price settles compared to AMD's offering.
8-core vs 8-core, even if AMD wins on price/performance, they don't win as much as they need to.
12-core vs 8-core, they do.

 

[turtile]

Remember that AMD assumed it would be competing against Intel's new 10nm process and new architecture when they designed this generation. The 'leaked' prices make sense if AMD needed to be competitive. Quickly releasing a whole new line will eat into AMD's current inventory and will lower margins. I highly doubt anyone within AMD knows how they will price the next generation 5 months out.

I think AMD will release high price processors first (Threadripper, then 8, 12 and 16 core). And once the old stock runs out, release the low end CPUs.

 

[exquisitechar]

Isn't that exactly why they DO need to release the 12-core?
Matching the 9900k is nice, but once Intel drops prices, it's 50/50 which you choose, and that favors Intel.
If the R7 is 12-cores and performs +30-50% above the 9900k, boom, AMD wins, and it's not even close.



The 9900k won't stay over $500, so the pertinent number is where the price settles compared to AMD's offering.
8-core vs 8-core, even if AMD wins on price/performance, they don't win as much as they need to.
12-core vs 8-core, they do.

Oh, I agree, I just misunderstood your original post and thought you meant that they wouldn't release a 12 core at launch, so I said that would be a mistake.

 

[PeterScott]

It must be a sort of Stockholm syndrome effect in play. Intel hijacks your wallet and we sympathize with their pricing.

Maybe, just maybe, AMD prices won't be basement ones. They just won't be sky-high ones. Using the present Intel as the lords of reasonable pricing is astounding.

It's not a questions of "sympathizing", it's recognizing how business actually works instead of using wishful thinking to imagine giving you great deals, just because it would be nice.

AMD, like every other company will want to maximize profit.

 

[mattiasnyc]

It must be a sort of Stockholm syndrome effect in play. Intel hijacks your wallet and we sympathize with their pricing.

Maybe, just maybe, AMD prices won't be basement ones. They just won't be sky-high ones. Using the present Intel as the lords of reasonable pricing is astounding.

What's this "sympathize with"?

Nobody is saying they do that or "like" Intel's pricing. This is capitalism. All we're doing is describing how it works and what is reasonable from the standpoint of the business. Every consumer wants more for less. Duh. Every business should want to maximize profits over X years. Apparently that's not as "duh"...

 

[mattiasnyc]

Who cares about the stock price? Look at the market share and revenue, AMD still has a long way to go.

If revenue and cash flow ever becomes an issue for a company then the stock price is really important, and more so its trajectory. If AMD needs more investment into it then that stock price is crucially important.

Yes, AMD has a lower market share and is trying to increase it. But financially it's doing pretty well and it'll be important moving forward not to get into an actual price war with Intel because that's going to not only damage revenue but also stock prices. It'd be a bad thing.