News AMD's Upd Ryzen 3 1200 With 12nm Zen+ Architecture Hits Retailers

[Admin]

AMD's three-year old Ryzen 3 1200 CPU gets a nifty upgrade.

AMD's Upd Ryzen 3 1200 With 12nm Zen+ Architecture Hits Retailers : Read more

 

[jeremyj_83]

AMD's three-year old Ryzen 3 1200 CPU gets a nifty upgrade.

AMD's Upd Ryzen 3 1200 With 12nm Zen+ Architecture Hits Retailers : Read more

That could end up being a very good budget gaming system CPU.

 

[King_V]

Well, if the 1600 AF is regularly available for $85, I wonder what the 1200 AF would be?

And, what does this do to the Athlon prices (although those have integrated GPU, so not exactly the same thing, but still, likely to have an effect).

 

[jeremyj_83]

Well, if the 1600 AF is regularly available for $85, I wonder what the 1200 AF would be?

And, what does this do to the Athlon prices (although those have integrated GPU, so not exactly the same thing, but still, likely to have an effect).

Does pose an interesting question in regards to the Athlon. The 3000G is currently the cheapest one at $60. If you need a dGPU & the Athlon 3000G & Zen+ R3 1200 are the same price, you go with the R3 every day. Really the top end Athlon is going to have to top out a $50 and only that much because it has the iGPU.

 

[King_V]

The Athlon prices are really weird. The 3000g of course undercut all the previous Athlons. Yet, somehow, despite being officially released at $49.99, the 3000g is generally selling for more (exception: MicroCenter in-store currently $44.99), and the old 2x0ge models are selling for equal to or more than the 3000g, despite the fact that they are all lesser performing.

This is not just because of the pandemic, either, it's been like that for a while.

I feel a Rick James meme about drugs seems applicable to Athlon pricing. I have a bit of a suspicion that even with the Ryzen 3 1200 AF and the release of the Ryzen 3 Zen+ CPUs, that the Athlon pricing is still going to be strange.

 

[InvalidError]

Wish AMD would just drop the MSRPs on 2000-series instead of re-badging them as 1000-serie and introducing unnecessary confusion.

 

[King_V]

I think maybe they don't guarantee that it hits the clock rates of the 2000 series? Not really sure.

There's got to be SOME reason for them to be putting this extra effort in. I hope.

 

[AnimeMania]

AMD should have spread the trace out on the old die size area to improve thermals and increase the GHz. This would have instantly become the low cost gaming solution.

 

[InvalidError]

I think maybe they don't guarantee that it hits the clock rates of the 2000 series? Not really sure.

The 1600AF could have been branded something like 2600LE or 2590. Having first-gen branding on second-gen is very screwy. For example, people who have OEM systems from companies like Dell and think they can put a 1600AF in get screwed over because their first-gen OEM motherboard does not support 2nd-gen chips.

 

[bit_user]

AMD should have spread the trace out on the old die size area to improve thermals and increase the GHz. This would have instantly become the low cost gaming solution.

Isn't that basically the definition of Zen+? They didn't change the layout - just the space in between the traces. I know that's oversimplified, but this point was emphasized in the initial reviews.

 

[alextheblue]

For example, people who have OEM systems from companies like Dell and think they can put a 1600AF in get screwed over because their first-gen OEM motherboard does not support 2nd-gen chips.

To be fair, I'm pretty sure the big OEMs will at least have Zen+ capable BIOS available.

 

[AnimeMania]

Isn't that basically the definition of Zen+? They didn't change the layout - just the space in between the traces. I know that's oversimplified, but this point was emphasized in the initial reviews.

I figured they shrunk everything down so they could get more chips from a single wafer, thereby reducing costs. I am disappointed that the shrinkage didn't yield higher clock speeds. All the specs seem the same. Seems like a waste of resources.

 

[InvalidError]

To be fair, I'm pretty sure the big OEMs will at least have Zen+ capable BIOS available.

Dell does not make BIOS updates for anything beyond whatever configurations were planned at launch since they do not support user mods to shipped configurations.

 

[cryoburner]

I am disappointed that the shrinkage didn't yield higher clock speeds. All the specs seem the same. Seems like a waste of resources.

These aren't exactly "new" processors. The 1600 AF is pretty much just a slightly underclocked 2600, and the 1200 AF is a lower-clocked 2300X (an OEM part), both from 2018. The main attraction here is the significantly lower pricing.

And technically, the listed clock rates are not the whole story, as only the base clocks (mostly meaningless for a desktop processor with adequate cooling) and the maximum boost clocks are listed. What they don't list is a the all-core boost clocks, and while I haven't seen benchmarks for the 1200 AF yet, the 1600 AF has been shown to maintain higher boost clocks when multiple cores are loaded, in addition to the slight IPC advantages of Zen+, resulting in performance that's generally rather close to a 2600 and slightly above a 1600X, let alone the original 1600, despite the official clock rates staying the same. I would expect the same might hold true for for the 1200 AF as well. And of course, Zen+ tends to overclock a little higher than the original Zen.

 

[TJ Hooker]

the 1600 AF has been shown to maintain higher boost clocks when multiple cores are loaded, in addition to the slight IPC advantages of Zen+, resulting in performance that's generally rather close to a 2600 and slightly above a 1600X, let alone the original 1600, despite the official clock rates staying the same.

That's a good point. The original precision boost kind of sucked IIRC, i.e. is anything more than two cores (or maybe just two threads) were loaded it more or less dropped down to base clocks. PB2 introduced with Ryzen 2k/Zen+ did a much better job of scaling boost with # of threads loaded, if the AF parts have PB2 they should have noticeably better stock performance that their 1st gen equivalents.

 

[punkncat]

I am curious.

If you install an "AF" variant of these onto a B350 chipset, does it natively see it, or would you need the BIOS update from the original sku (1st gen)?

 

[InvalidError]

If you install an "AF" variant of these onto a B350 chipset, does it natively see it, or would you need the BIOS update from the original sku (1st gen)?

No, the 1xxx AF chips are 2nd-gen and require a 2nd-gen aware BIOS. If you have one of those boards that will never get such a BIOS update, you are out of luck.

And even on boards with 2nd-gen BIOS, you may still require a post-AF BIOS update if the BIOS checks the installed chip against a list of known models to determine support status.

 

[alextheblue]

Dell does not make BIOS updates for anything beyond whatever configurations were planned at launch since they do not support user mods to shipped configurations.

Well that's pretty awful, especially given they do utilize newer microcode in their BIOS updates, so they're really artificially gimping upgrade opportunities. Another reason to avoid Dell. I certainly hope that isn't the norm for OEMs these days, in the past I've had good luck with upgrading OEM machines. Though perhaps they were less prone to new mainboard revisions in those days.

 

[bit_user]

Dell does not make BIOS updates for anything beyond whatever configurations were planned at launch since they do not support user mods to shipped configurations.

FWIW, I installed a Broadwell-EP Xeon in a Dell Precision workstation that originally shipped with a Haswell-EP. I upgraded the BIOS before the swap, so I can't say whether it would've worked otherwise.

However, the distinction might be that Dell went on to offer said Broadwell CPUs in that platform, whereas their original Ryzen 3 systems might've been discontinued rather than shipping with these new CPUs.

I do wonder why we're worried about someone being unable to install a new Ryzen 3 CPU in a legacy Ryzen 3 system... what would be the use case for that?

 

[bit_user]

No, the 1xxx AF chips are 2nd-gen and require a 2nd-gen aware BIOS. If you have one of those boards that will never get such a BIOS update, you are out of luck.

With regard to B350 boards getting BIOS updates, @drea.drechsler recently confirmed that some such boards could get updates enabling them to run Zen+ CPUs:

https://forums.tomshardware.com/thr...-for-you-when-upgrading.3590695/post-21684592​

 

[InvalidError]

I do wonder why we're worried about someone being unable to install a new Ryzen 3 CPU in a legacy Ryzen 3 system... what would be the use case for that?

Because "new" CPUs labeled as 1000-series become problematic when people with "legacy" boards that will never have official support for newer CPUs "mysteriously" fail to work with some 1000-series SKUs.

Shouldn't be so hard to imagine how someone who may have a Ryzen 1200-1400 pre-built could be mighty miffed that a "1600" does not work on a board that should support anything at least up to the 1700. Without the additional prior knowledge, it would be very easy to overlook the "AF" (in general, CPU model suffixes don't make or break compatibility within a given socket type) or expect the parts to be directly interchangeable since the first digit is supposed to represent the generation.

 

[bit_user]

Because "new" CPUs labeled as 1000-series become problematic when people with "legacy" boards that will never have official support for newer CPUs "mysteriously" fail to work with some 1000-series SKUs.

Shouldn't be so hard to imagine how someone who may have a Ryzen 1200-1400 pre-built could be mighty miffed that a "1600" does not work on a board that should support anything at least up to the 1700.

When I posted that, I was only aware of the 1200 refresh and wondering why anyone would even bother to "upgrade" to that. If I'd been more specific in my question, that might've been clear.

 

[InvalidError]

Whoa there! Might want to check your facts, before so confidently making such blanket assertions.

You misread what I wrote. "One of those [b350] boards that will never get an update" does not mean that no b350 boards whatsoever will ever get an update, it means that some never will. While most DIY boards did get updates, albeit at the expense of backward compatibility or features in some cases, large OEMs with custom motherboard and BIOSes (ex.: Dell) rarely bother updating BIOSes to support anything beyond what they expect to ship systems with.

 

[bit_user]

You misread what I wrote. "One of those [b350] boards that will never get an update"

True.

Sorry for calling you out, but anyway the link I posted might help some. Edited.

 

[drea.drechsler]

With regard to B350 boards getting BIOS updates, @drea.drechsler recently confirmed that some such boards could get updates enabling them to run Zen+ CPUs:
....

I can't claim authority on that, but it's really elementary and shouldn't be a mystery. Anyone with a B350 board who wants to truly confirm if their board is able to run Zen+, Zen2 or the 'AF' hybrid CPU's should just check the manufacturer's support web page for their board. All of them list CPU compatibility and at the BIOS ver. where it starts.

I hate to be unnecessarily inclusive as that can be dangerous, but support has extended to later BIOS releases in every case I've been seen. There's almost always an exception though, so that's why checking BIOS release notes is always important.

I'm going on a limb, but it will most likely be the same when Zen2+ (Zen3? or whatever they end up calling it) comes out for all currently fielded AM4 boards.