AMD May Release a Ryzen 7 2800X CPU “Someday”

Megalith

24-bit/48kHz
Staff member
Joined
Aug 20, 2006
Messages
13,000
According to Senior Vice President and General Manager Jim Anderson, AMD did not bother releasing a 2800X because the recent Ryzen lineup (2600X and 2700X) already covers all current performance and price points. He does admit that a 2800X is not out of the question, which has led some to think that it will be used to counter Intel’s rumored 8C/16T CPU.

“We felt like, with the 2700X and 2700 at the performance and price points, we had that space covered. We just felt that with those two SKUs we had it sufficiently covered where we wanted to position the product. That doesn’t preclude a 2800X someday, right, maybe. But for now we believe those two SKUs cover the space well.”
 
Most reviews have shown that you have to push pretty hard to get any faster than a 2700X can do at stock (with upgraded cooling), which leads to the supposition that AMD literally can't make anything faster. Yet.

Maybe they have a 'Zen++' stepping in the pipe but don't want to do another retail incrementation of the same arch, so they left that established product spot open?
 
Kudos to them. Not like the Intel method of throwing **it at the fan and hoping something sticks. Hoping they are saving the higher bines for TR2 or that.
Edit- Save another round, at least one for yourself.
 
2800X with a base clock of 4.0 Ghz and XFR around 4.5????

One can dream....
 
Kudos to them. Not like the Intel method of throwing **it at the fan and hoping something sticks. Hoping they are saving the higher bines for TR2 or that.
Edit- Save another round, at least one for yourself.

Based on the last Ryzen launch, they're pretty much selling everything they can produce right now and in this case a better bin would only yeild like 100mhz, which isn't really a different product.. As the process matures, and they can refine their current tech, it wouldn't be surprising to see a few more mhz at a similar power envelope possible, but until Intel has something that drastically changes the market shape, this is all moot.
 
1800x was a mistake, or at least the 1700x shouldn't have existed, i can't count youtubers and reviewers that were puting ryzen 7 at 500$ just because of the 1800X, although the exact same cpu is at 300$
stock + overclocked variant seem enough, what's the point of puting an average overclock version in the middle, just stupid.
beside what could AMD do with a 2800X nothing beside overclock, unless they found a way to squize in a 3rd CCX in the package with 12nm, which i doubt.
 
Edit: I see the quote now. Some other websites never included that which irked me a bit. Amazing how rumors can be like an old game of telephone sometimes.

All the conjecture I believe stems from a PC World video interview of Jim Anderson covering the Ryzen 2nd gen chips.

In that interview he was asked about the possibility of of a 2800x chip. He said essentially in their eyes the 2700x is the replacement of what the 1800x was.

However when pressed on the future release of a 2800x chip, he corporately neither confirmed nor denied the possibility of such a chip ever being released.

So basicly this is a nothing sandwich. It details nothing more than already could be speculated.

Fun to talk about though. People seem to think a core increase wouldn't happen as current CCX designs contain 4 cores each. So a 5 core ccx might not work. And a 6 core CCX would interfere with Threadripper SKUs and might not fit so well on AM4 I'd imagine.
 
Last edited:
If they do release a 2800X and it follows the same path as the 1700X/1800X then is it the same CPU clocked faster and seeing as the 2700X can be overclocked to 4.3ghz, there would be nothing to be gained by buying a 2800X.
 
1800x was a mistake, or at least the 1700x shouldn't have existed, i can't count youtubers and reviewers that were puting ryzen 7 at 500$ just because of the 1800X, although the exact same cpu is at 300$
stock + overclocked variant seem enough, what's the point of puting an average overclock version in the middle, just stupid.
beside what could AMD do with a 2800X nothing beside overclock, unless they found a way to squize in a 3rd CCX in the package with 12nm, which i doubt.

Well, the 1800x at US$500 was as fast as you could get out of the box- and it gave AMD a means to gauge price ceilings for their new part. As the pricing for Ryzen 2 shows, they seem to have shot a little high.

If they do release a 2800X and it follows the same path as the 1700X/1800X then is it the same CPU clocked faster and seeing as the 2700X can be overclocked to 4.3ghz, there would be nothing to be gained by buying a 2800X.

It'd have to go higher, thus the speculation of better binning or potentially a new stepping in the pipe that isn't ready yet. This makes further sense in the context that 4.3GHz seems to be an architecture limitation, and if they could get boosts to 4.5GHz, would certainly be worthy of another SKU (and higher prices with potentially larger profit margins).

Edit: I see the quote now. Some other websites never included that which irked me a bit. Amazing how rumors can be like an old game of telephone sometimes.

All the conjecture I believe stems from a PC World video interview of Jim Anderson covering the Ryzen 2nd gen chips.

In that interview he was asked about the possibility of of a 2800x chip. He said essentially in their eyes the 2700x is the replacement of what the 1800x was.

However when pressed on the future release of a 2800x chip, he corporately neither confirmed nor denied the possibility of such a chip ever being released.

So basicly this is a nothing sandwich. It details nothing more than already could be speculated.

Fun to talk about though. People seem to think a core increase wouldn't happen as current CCX designs contain 4 cores each. So a 5 core ccx might not work. And a 6 core CCX would interfere with Threadripper SKUs and might not fit so well on AM4 I'd imagine.

First, I think AMD has found themselves in a catch-22 with the *800x SKU: if they released their top silicon as the '2800x' instead of the 2700x at $329, they'd be seen as significantly cutting retail prices. It's also clear that the 2700x is doing everything that the 1800x did, so now the question as to why that SKU position wasn't also moved forward arises.

Second, with the drop to 7nm (or whatever), AMD should be able to build a six-core CCX. Assume that this is coming with DDR5 and increased bandwidth per channel, and they could easily produce a 12-core R7 for 'AM5'. They might continue to produce quad-core CCXs as well, though.
 
Hopefully this gives them more room fore TR4 SKUs.

2800X will be a quad channel 4.0-4.3Ghz 8-core part.
 
Man I'll just be happy someday when there is a 12 core ryzen part at $400 and under.

If only because I want to see the resale value of 6950x chips on ebay become utterly oblitered so I can grab one to replace my 6800k (Probably never going to happen, but a man can dream).
 
Most reviews have shown that you have to push pretty hard to get any faster than a 2700X can do at stock (with upgraded cooling), which leads to the supposition that AMD literally can't make anything faster. Yet.

I wouldn't be so quick to discard this, they've got a die shrinkage that has not been utilised as of yet; they literally left empty silicone where transistors once were (check die size, waffers, check pics that have circulated, think it was on Anandtech). On top of that, the issue with clocking these things relies mostly on complex-to-complex communication energy costs; Infinity Fabric eats up significant power, i'd remind here they had to redo the entire controller stepping to bring the EPYCs in line. And also, that what Intel considers core and uncore are here interlinked. One thing having an Infinity Fabric with an Intel uncore that clocks lower; quite another when the IF needs facilitate an uncore clocked at 4.0, 4.1 or 4.2GHz.
If a proper utilisation was made (because currently it's only cost saving) and if an allowance for a significantly higher TDP was considered 'O.K.', i could definitely see them producing a chip that clocked around 150, 200 more than a 2700X does.
And it's not something they've shied away from doing before; the entire FX 9xxx series was just ridiculously overclocked 8xxxs.

Would i buy something with a significantly higher TDP? Personally, sure. Would others though? Can see why they'd hesitate. Whereas, if Intel wipes the floor with the new ones..

To the user above me wishing for a 12 core Ryzen so he can buy a 5 years old Intel CPU (by that time), dude, lol, that's some pretty forward thinking you've got there.
 
Last edited:
Good points- though one of the issues here is that such a part would have severe motherboard limitations; if it's not a whole new line (like Intel did with the 8th gen) then they'll have to be very careful about motherboards.
 
  • Like
Reactions: Aenra
like this
I wouldn't be so quick to discard this, they've got a die shrinkage that has not been utilised as of yet; they literally left empty silicone where transistors once were (check die size, waffers, check pics that have circulated, think it was on Anandtech). On top of that, the issue with clocking these things relies mostly on complex-to-complex communication energy costs; Infinity Fabric eats up significant power, i'd remind here they had to redo the entire controller stepping to bring the EPYCs in line.
If a proper utilisation was made (because currently it is only cost saving) and if an allowance for a significantly higher DTP was considered 'O.K.'. i could definitely see them producing a chip that clocked artound 150,200 more than a 2700X does.
And it's not something they've shied away from doing before; the entire Fx9 series was just ridiculously overclocked 8xxxs.

Would i buy something with significantly higher TDP? Personally, sure. Would others though? Can see why they'd hesitate. Whereas, if Intel wipes the floor with the new ones..
Interesting about the silicone filler. I was curious after der8auer delided a 2600x and found the die was the same size as it was in the past despite the 12nm shrink.

Could just be a design choice, but it would be neat to see if they have any plans there!
 
I was curious

They have the yields all set by now, this is money earned with no extra effort; on top of which, the dead zone acts as a secondary heat buffer, which in turn explains the lower than expected TDP. Which is why PB 2.0 looks so good right now in terms of performance-to-cost.
 
Good points- though one of the issues here is that such a part would have severe motherboard limitations; if it's not a whole new line (like Intel did with the 8th gen) then they'll have to be very careful about motherboards.

Do new folks get banned for multi-posting? :)
I'd edit and merge, but if i recall correctly, Xenforo messes up the notifications when you do that. My apologies if this is considered unseemly here.

In regard to motherboards, i know for a fact that the new Gigabyte 7 and the 470 Taichis have a near 100% efficiency pumping 125 Amps on the CPU. That's enough to accommodate more than is currently required.
However we agree, as i can also recall Asus' Crosshair.. that junk could not run a 9590 properly for its life; i personally had to get me an AsRock Extreme 9 to run it to its fullest.
So yeah, this could be an issue assuming an uninformed audience. Too far down the line though ^^
 
In regard to motherboards, i know for a fact that the new Gigabyte 7 and the 470 Taichis have a near 100% efficiency pumping 125 Amps on the CPU. That's enough to accommodate more than is currently required.
However we agree, as i can also recall Asus' Crosshair.. that junk could not run a 9590 properly for its life; so yeah, this could be an issue assuming an uninformed audience.

That would have been the problem Intel would have had, and the problem that AMD did have- you can have new boards marked, but power delivery just isn't that high on the spec sheet and these new CPUs will be sold at retail right next to heaps of boards that at best will not power them properly, or at worst, fail and break when stressed. Given that this would be an upjuiced part, AMD might be better off adding a chipset SKU just for it, but allow for OEMs to mark boards with the lower chipsets that can handle it.
 
is there enough headroom left in Ryzen+ to release a competitive part to counter intel in the future? Call me skeptical.
 
Given that this would be an upjuiced part, AMD might be better off adding a chipset SKU just for it, but allow for OEMs to mark boards with the lower chipsets that can handle it.

Yeah, agreed.
Which leads us to the 490 chipset rumor that was allowed..? Enouraged..? To be recently circulated ^^

* To hondaman above, they already are competitive; if you haven't been reading the news, you may as well start. Unless.. To paraphrase Google, did you mean "bestest"? Because i'd then have to ask what it is that you do with your CPUs, as statistically speaking at least, i sure as hell don't get all the excitement about an L3-stripped CPU that can clock at 5+. Of course it can, now that they stripped its L3. There's a reason why my 6900K could at best reach 4.5 and all of sudden the next 8cores go to 5.2Gigs ^^
 
Last edited:
Yeah, agreed.
Which leads us to the 490 chipset rumor that was allowed..? Enouraged..? To be recently circulated ^^

* To hondaman above, they already are competitive; if you haven't been reading the news, you may as well start. To paraphrase Google, did you mean "bestest"? Because i'd then have to ask what it is that you do with your CPUs, as statistically speaking at least, i sure as hell don't get all the excitement about an L3-stripped CPU that can clock at 5+.. Of course it can.. not that they stripped its L3..
You got me entirely wrong. I own 3 Ryzen boxes right now. I've only bought AMD stuff for 20 years so calm down!

What I'm saying is (again) that there doesn't seem to be much headroom left in Ryzen to compete with a FUTURE Intel "go-fast" part.
 
is there enough headroom left in Ryzen+ to release a competitive part to counter intel in the future? Call me skeptical.

Assuming Intel drops an eight-core i7 into their consumer socket, with graphics and clockspeed behavior mirroring the 8700k, then no, AMD doesn't have the headroom without at least a new stepping, if what we've seen from Zen+ holds.

Of course, if Intel releases said part in the US$500 range, then the situation is fairly moot. If they put it in the US$350 range, i.e. taking over for the 8700k, then AMD is going to have to drop prices and bide their time to Zen2.

Which will hopefully arrive before Intel's next architecture.
 
Does anyone know of the bug or issue of reduced 1080p gaming performance is still and issue with Ryzen? Would love to build an AMD PC again.
 
Does anyone know of the bug or issue of reduced 1080p gaming performance is still and issue with Ryzen? Would love to build an AMD PC again.

It's not 'reduced'- it's just slower. You'll need to be pushing your system for it to matter, though, like with any app.
 
calm down!

Not an issue; i often tend to point out how people's understanding of the term 'competitive' is either flawed to begin with, or worse, entails some degree of a placebo effect.
If your understanding of the term differs and/or is actually the literal one, your post did not make see it.

No offense meant either way, if it was my typing "bestest" that prompted this, mere exasperation. I simply find the way we perceive all of this a touch too narrow-minded, the usual one word buzz term that praises or buries a product and that's that kind of a thing; but that's just my personal opinion :)
 
They have the yields all set by now, this is money earned with no extra effort; on top of which, the dead zone acts as a secondary heat buffer, which in turn explains the lower than expected TDP. Which is why PB 2.0 looks so good right now in terms of performance-to-cost.

As of today the yields may be set, however the silicon being sold today is coming from yields a couple of months ago. They are, most likely, already holding back the dice that are coming in at higher efficiency levels with less leakage and potential for higher clocks.

The question is at what levels will these exceptional dice perform at. 4.1 with 4.4 boost? Regardless, this 2800s won't be high volume sellers which means the best dice will start being held back for Threadripper's launch in the Fall.
 
It's not 'reduced'- it's just slower. You'll need to be pushing your system for it to matter, though, like with any app.

Awesome, I have no interest right now to upgrade from my plasma TV. Maybe in due time but 1080p life for me.
 
Awesome, I have no interest right now to upgrade from my plasma TV. Maybe in due time but 1080p life for me.

The bottleneck that ryzen introduces is usually at 120-160 FPS.

so MUCH faster than your plasma can handle. You'll be fine.
 
I think they wanted the best dies for threadripper rather than releasing them in a 2800x
 
Yeah I'd say AMD is holding out so as not to cannibalize current thread ripper sales while they still command a larger profit margin until they receive the same refresh treatment.
 
I'm curious what all-core clock speeds are listed by SiliconLottery this week (April 28th date is looming). Particularly for the Ryzen 2000-series, it makes even less sense to be paying a premium, even if it's a small one ($20 or $40). What's the benefit - maybe another 100-150Mhz over the typical 4.2Ghz we're seeing?

I bring this up b/c there's going to be an inherent expectation (deserved or not) that the 2800X guarantees better results. We know from the 1800X that it was not in the past :p
 
If they do release a 2800X and it follows the same path as the 1700X/1800X then is it the same CPU clocked faster and seeing as the 2700X can be overclocked to 4.3ghz, there would be nothing to be gained by buying a 2800X.
Most people are most certainly not getting 4.3 out of it.
 
I think AMD would be wise to design a 12-core AM4 chip for Ryzen 3. We learned with Raven Ridge that CCX =/= Memory channel, so having dual channel chips with a 3 CCX design is by no means science fiction.

I highly doubt it would be any time soon. Mid 2019 by the earliest, but if Intel has the balls to pump out an 8-core 16-thread 115X chip with integrated video, then you're looking at a pretty tough time for AMD if they stick to their dual CCX design on their silicon.

The other option is to add 2 cores to the Zen 2 (Ryzen 3000) CCX. That way a dual-CCX silicon would have 12 cores, 24 threads, and a single CCX chip (I-E APUs) would have 6/12.

I trust Zen 2 will be a pretty big jump in IPC, but it won't match Intel, and likely won't clock as far, so I doubt remaining on an 8/16 mainstream silicon will be a wise idea to maintain a compute lead. Having an 8/16 CPU on the mainstream will stop becoming a novelty sooner rather than later if Intel hops in the fray. AMD needs to do something to stay ahead in Raw compute.
 
JUST BUILD THE DAMN THING YOU IDIOTS!


The problem is that there isn't much room left to go with the silicon.

The 2700X already effectively can't overclock, as any stable speed is actually an underclock. I doubt AMD can actually release a faster chip with the silicon available.
 
The problem is that there isn't much room left to go with the silicon.

The 2700X already effectively can't overclock, as any stable speed is actually an underclock. I doubt AMD can actually release a faster chip with the silicon available.

That's true I suppose. Bring on the 7nm, ladeez!

If Intel brought out a new arch with, say, +15% IPC improvements, AMD will be hurting again.
 
That's true I suppose. Bring on the 7nm, ladeez!

If Intel brought out a new arch with, say, +15% IPC improvements, AMD will be hurting again.

I agree, but I don't see that happening with our third year on the same architecture with Intel. I think AMD is posed for a 15% IPC upgrade more than Intel.
 
That's true I suppose. Bring on the 7nm, ladeez!

If Intel brought out a new arch with, say, +15% IPC improvements, AMD will be hurting again.

Not quite as badly, as the gaming bottlenecks are much less with Ryzen as compared to Bulldozer. Besides, from Sandy Bridge to Skylake has been an aggregate of approximately 15-20% IPC, and you have people still saying Sandy Bridge is good enough. A 15% jump isn't going to mean as much in the current market.

That said, a jump like the C2 to C3 stepping of the Phenom II era would be nice to see.
 
Back
Top