AMD: Ryzen 5000 B2 revision CPUs won't offer performance improvements

jsilva

Posts: 325   +2
TL;DR: AMD has come out publicly to confirm the recently spotted B2-stepping Ryzen processors indeed exist, but they won't bring any significant performance improvements over their predecessors. According to an AMD representative, the new B2 revision is the result of an enhanced manufacturing process and should be available over the next six months.

The leak suggesting that AMD was possibly launching the Ryzen 5000XT series as B2 stepping CPUs included two SKUs: a 16-core CPU with a 3.4GHz base clock and a 5.0GHz boost clock, and a 6-core part with a 3.7GHz base clock and a 4.6GHz boost clock.

After being questioned about this development, AMD made a statement confirming that these are not new processors. The new B2 stepping processors are just the same as the Ryzen 5000 chips currently available, which are on the B0 stepping. The only difference is that they result from a new manufacturing process.

Also read: Top 5 Best CPUs

"In continuous efforts to enhance our manufacturing and logistics capabilities, AMD is gradually transitioning the AMD Ryzen 5000 Series desktop processors to a 'B2' revision over the next 6 months. There are no feature, function, or performance enhancements to the B2 revision, and no BIOS update is required," said an AMD representative.

It's very common for chip makers to release new product revisions when replacing some components, improving processes, squashing bugs, or some others. Although AMD hasn't confirmed the reason to create a new revision, it's possible that the B2 CPUs will come with slight adjustments to improve their yields, therefore increasing the Ryzen 5000 series production and stock.

Another plausible explanation for a new Ryzen 5000 revision would be to fix the PSF vulnerabilities introduced in the Zen3 chips and/or the Spectre vulnerabilities found in the micro-op cache of Zen-based processors. Bottom line, most end-users won't be able to distinguish a Ryzen 5000 B0 CPU from a B2 CPU.

Permalink to story.

 
Well, I don't think Intel will offer anything that comes near in 2020 anyway. It's not a terrible strategy to just cruise on first place without spending more gas than you need to speeding up if your opponent isn't catching up at all.
 
Well, I don't think Intel will offer anything that comes near in 2020 anyway. It's not a terrible strategy to just cruise on first place without spending more gas than you need to speeding up if your opponent isn't catching up at all.

That's what Intel did for so many years.
 
I'm still waiting for 5700(x). There is a huge gap between 5600x and 5800x.
What sorts of gap? 5600X = 6c/12t, and 5800X = 8c/16t. Looks normal to me. An in between solution sounds pointless to me. If you are expecting a 3700X vs 3800X situation, I think that may not happen. Its obvious to AMD that 3700X basically negates any point for a 3800X, similar to the current Intel Rocket Lake 11700K vs 11900K. The difference is really marginal, and unless you need every single Mhz that you can squeeze out of it, there is no point paying a premium for a similarly specced but marginally faster out of the box performance solution.
 
Well, I don't think Intel will offer anything that comes near in 2020 anyway. It's not a terrible strategy to just cruise on first place without spending more gas than you need to speeding up if your opponent isn't catching up at all.
Intel offers good alternatives to AMD in 11600, 11400 and the last-gen parts. Intel competes with AMD more than AMD competed with Intel pre-Ryzen.

I'm still waiting for 5700(x). There is a huge gap between 5600x and 5800x.
There's a gap between 5600X and...well, there's nothing below it.
 
Intel offers good alternatives to AMD in 11600, 11400 and the last-gen parts. Intel competes with AMD more than AMD competed with Intel pre-Ryzen.

There's a gap between 5600X and...well, there's nothing below it.

Agreed, I think the problem for AMD is that a sub 5600X doesn't make sense from a manufacturing perspective. The yields on the CCD cores are likely so high that they generate insufficient dies that have less than six functional cores or reach 5600X clock speeds to warrant a product line, and even if they spun out a 4 core or 6 core with less cache, the fact they still need the same IO die means the manufacturing cost saving is negligible. Intel on the other hand has fabs to fill, any less than 100% utilisation hurts profit margins greatly, so they are willing to sacrifice ASP to keep their fabs fully utilised. AMD only needs to fulfill its GF and TSMC order obligations which are lower than demand. Which is why supply for the older 3000 series is drying up, why waste limited 7nm on older cheaper products.

I do wish AMD taped out a cut down Renoir design with 4 core and Vega 4-6, likely would be sub 80mm2 and would make a great cheap APU with good availability, especially if it remained sub 25W TDP. Not everything needs 6-8 core.
 
What sorts of gap? 5600X = 6c/12t, and 5800X = 8c/16t. Looks normal to me. An in between solution sounds pointless to me. If you are expecting a 3700X vs 3800X situation, I think that may not happen. Its obvious to AMD that 3700X basically negates any point for a 3800X, similar to the current Intel Rocket Lake 11700K vs 11900K. The difference is really marginal, and unless you need every single Mhz that you can squeeze out of it, there is no point paying a premium for a similarly specced but marginally faster out of the box performance solution.
Obviously we need a 7C/14T
 
That's what Intel did for so many years.
Intel didn't stop trying to bring out new technologies their fabs just couldn't keep up with their designs. AMD's old fab is still making chips just as dense as Intel's 7700K processors. It seems designing CPU's on a smaller process is much easier than actually making them on a smaller process.
 
Intel offers good alternatives to AMD in 11600, 11400 and the last-gen parts. Intel competes with AMD more than AMD competed with Intel pre-Ryzen.

Fair enough those are good points. There are very notable quirks though: even on those lower end parts the power requirement is such that it penalizes performance without a capable motherboard, killing the price parity. Technically intel *can* match AMD on performance and also on price if we only take the CPU price but this is something that hasn't improved in intel for years: motherboard choice is just inherently more costly.

But hey I know of at least one use case in which I'd take the extra motherboard requirements: If you are building an itx rig and also need to have integrated graphics in addition to a GPU then intel it's still a better idea until the 5600g and 5700g become more readily available.
 
But hey I know of at least one use case in which I'd take the extra motherboard requirements: If you are building an itx rig and also need to have integrated graphics in addition to a GPU then intel it's still a better idea until the 5600g and 5700g become more readily available.
Why would you need integrated and dedicated GPU's at the same time?
 
Why would you need integrated and dedicated GPU's at the same time?

GPU Pass-through: most every day tasks could be done on Linux or a fully separate VM for work stuff and then another one with the GPU for gaming.

Also right now it would be useful to build a system without needing to get a Geforce 710 or some equally worthless potato GPU while we wait at least a few more months for some GPUs to have reasonable prices.

I'm hoping a possible 3050 ti would mitigate the issues (Though not without a possible, rumored limitation of only 4gb vram) but right now I need to build a second rig for someone else and have no GPUs to put in it/mine to at least get a usable desktop.
 
I'm waiting patiently to get my 5950x at MSRP. An XT variant would have been nice, but these CPUs are so fast, that honestly it doesn't matter at all, except maybe to get a small discount on the 5950x.
 
Using a new manufacturing process is neither free nor cheap. There had to be some reason for the switch. Maybe the B2 silicon can be produced faster to address the silicon shortage.
 
GPU Pass-through: most every day tasks could be done on Linux or a fully separate VM for work stuff and then another one with the GPU for gaming.
I mean, true, but why wouldn't you just use your GPU for both? Seems like a lot of effort otherwise.
 
I mean, true, but why wouldn't you just use your GPU for both? Seems like a lot of effort otherwise.
2 reasons:

1) I have no other GPU right now: just a 1070 and cannot buy anything else but maybe a 710 and for such a crap product I would much rather just have integrated graphics.

2) I also rather build an ITX system if I can help it, so that means is mostly just one pci-e slot. You could get a mobo with a secondary m.2 slot and connect a secondary GPU to those but the solutions for that are not pretty: lots of lose cables and probably no way to properly close tiny case once that is installed
 
Using a new manufacturing process is neither free nor cheap. There had to be some reason for the switch. Maybe the B2 silicon can be produced faster to address the silicon shortage.
TSMC already has at least 7, 7+ and 7NP. Perhaps Ryzens are now made with 7+ or 7np while previously used "initial" or "refined" 7. Possibly AMD got bigger share of those "better" 7nm processes and now has enough to make Ryzens on those. "New process" in this case means much less than changing from 16FF to 7nm for example.

I mean, true, but why wouldn't you just use your GPU for both? Seems like a lot of effort otherwise.
One GPU for two operating systems running at same time? Not possible. Oh yeah, it is "possible" but for other systems GPU runs very limited. I do exactly that and watching videos is mostly "OK" on limited OS.
2 reasons:

1) I have no other GPU right now: just a 1070 and cannot buy anything else but maybe a 710 and for such a crap product I would much rather just have integrated graphics.

2) I also rather build an ITX system if I can help it, so that means is mostly just one pci-e slot. You could get a mobo with a secondary m.2 slot and connect a secondary GPU to those but the solutions for that are not pretty: lots of lose cables and probably no way to properly close tiny case once that is installed
I think question was: why you don't use gaming GPU for both OS's.
 
One GPU for two operating systems running at same time? Not possible. Oh yeah, it is "possible" but for other systems GPU runs very limited. I do exactly that and watching videos is mostly "OK" on limited OS.
2 reasons:

1) I have no other GPU right now: just a 1070 and cannot buy anything else but maybe a 710 and for such a crap product I would much rather just have integrated graphics.

2) I also rather build an ITX system if I can help it, so that means is mostly just one pci-e slot. You could get a mobo with a secondary m.2 slot and connect a secondary GPU to those but the solutions for that are not pretty: lots of lose cables and probably no way to properly close tiny case once that is installed
Right, you still don't make any sense, if you're gaming, you're most likely using Windows 10, so that's your main OS, put Hyper-V on, install Linux, that's your "work" OS.

Or do it the other way round if you game on Linux, or duel boot if you don't need to run both at the same time.

I run plenty of VM's at home between my Microserver and my Desktop and I've never needed an iGPU to run a VM...
 
What sorts of gap? 5600X = 6c/12t, and 5800X = 8c/16t. Looks normal to me. An in between solution sounds pointless to me. If you are expecting a 3700X vs 3800X situation, I think that may not happen. Its obvious to AMD that 3700X basically negates any point for a 3800X, similar to the current Intel Rocket Lake 11700K vs 11900K. The difference is really marginal, and unless you need every single Mhz that you can squeeze out of it, there is no point paying a premium for a similarly specced but marginally faster out of the box performance solution.
Actually AMD already has two processors between 5600x and 5800x. 5800 and 5700G. Issue is both are OEM only.

As far as the gap, I'm sorry I meant to say price gap. 5600x & 5800x are like $A480 and $A700 respectively. That's like 220 dollar difference. Intel has 2 current gen cpus in that range (4 if you count 'F' variants). Surely, AMD could manage to put something in that price range?
 
I guess so, but you'd now have access to Nvidia NVENC encoder which is far superior.

Thing is not everyone has an nvidia GPU, some have AMD; also, not everyone has a "high" performance GPU, or high enough for their needs, so there are cases where even with an Nvidia GPU installed you still want to use Quick Sync (here examples). Then there are the cases where only the Intel solution helps, like happened time ago with some video editing software (was Adobe?), probably some other cases where only QS works or is supported...
Then imagine cases where a user needs to use those capabilities but has more than 1 work to do: in that case a second worker will help a lot. The simplest example I can imagine is someone working in editing a commercial, makes 2 versions with few differences, lets say in one case one scene is not there, but can't encode both one after the other and deliver on time...
Then software that can take advantage of both at the same time, not sure about this.
 
Back