Ryzen 1600X 6c12t 3.6/4.0 GHz $249 - the gaming CPU arrives 4/11

[DaveSimmons]

https://arstechnica.com/gadgets/201...ream-with-ryzen-5-4-cores-8-threads-from-169/

http://www.anandtech.com/show/11202/amd-announces-ryzen-5-april-11th

My main criticism of the 1800X as a gaming CPU is that it costs $150 more than the 7700K and is currently slower in games. If you don't have CPU heavy non-gaming tasks you pay more to get less (for now).

The 1600X makes a lot more sense. It's just as fast per core as the 1800X so it's probably going to be as fast for current games, it still has the 2 extra cores over the 7700K to possibly let it pull ahead in newer more highly threaded games, and it's $100 cheaper than the 7700K. It's also $80 cheaper than the 1700, and factory-tested to run without errors at 3.6/4.0 stock speed, no overclocking needed.

For anyone who bought into the "Ryzen is a fail for gaming" backlash, does this change your mind?

 

[richierich1212]

For anyone who bought into the "Ryzen is a fail for gaming" backlash, does this change your mind?

Anyone who believes Ryzen is a fail for gaming is trolling or is biased.

Just put a G4560 in every computer; who needs more than 2 cores/4 threads anywho.

 

[Crono]

The launch date is useful info, but the Ryzen gaming debate has already been hashed and rehashed numerous times in other threads. Until we get benchmarks and see real info on Ryzen 5 and 3 performance, it's not worth getting into the gaming debate yet again.

 

[Shivansps]

As 3+3? AMD is betting that 3+SMT + 3 to background tasks end up better than a 4/4, it may just be. But i whould wait for reviews, some really wierd scheduling stuff may happen there.

 

[frozentundra123456]

https://arstechnica.com/gadgets/201...ream-with-ryzen-5-4-cores-8-threads-from-169/

http://www.anandtech.com/show/11202/amd-announces-ryzen-5-april-11th

My main criticism of the 1800X as a gaming CPU is that it costs $100 more than the 7700K and is currently slower in games. If you don't have CPU heavy non-gaming tasks you pay more to get less (for now).

The 1600X makes a lot more sense. It's just as fast per core as the 1800X so it's probably going to be as fast for current games, it still has the 2 extra cores over the 7700K to possibly let it pull ahead in newer more highly threaded games, and it's $100 cheaper than the 7700K. It's also $80 cheaper than the 1700, and factory-tested to run without errors at 3.6/4.0 stock speed, no overclocking needed.

For anyone who bought into the "Ryzen is a fail for gaming" backlash, does this change your mind?

The so called "backlash" is probably a reaction to the hype and attempts by a few AMD fans in these forums to promote the 1800 as the best gaming cpu, but whatever. It wont change my opinion, but it never was my opinion that Zen was a "fail" for gaming, just not the best choice at the current time. The big question is, I think, if the quad and hex cores overclock better than the 8 cores, but I am not expecting it too, since it appears the more expensive cpus are better binned than the cheaper ones. I probably would pick it over a 4/4 kaby lake, but it would depend on the budget if I would pick it over a 7700k.

 

[SPBHM]

quad core is 2+2, hmmm, I was hoping for a just a single ccx
interesting that one quad core is with 16MB and the other 8MB l3, looking at these specs I assumed the 8MB could be a single ccx, but by the article it sounds like they are all 2+2.

 

[draizze]

Ok now that R5 has been confirmed as 3+3 (of course) and 2+2 (a little bit surprise), can we have R7 benchmark with disabled core configuration please?

 

[Triloby]

Interesting. The R5 1500X looks to have the highest XFR out of all of them. An extra 200 MHz on top of the Turbo frequency, but maybe that only applies for a single core and thread? Either way, it looks like that specific SKU has the potential to overclock better.

It also makes me wonder what the point of the 1400 is considering it's only $20 cheaper than the 1500X, yet it has less L3 cache and XFR boosts. Maybe the lesser L3 cache won't matter much in games, but the $20 difference makes it seem pointless in the face of the 1500X.

 

[guachi]

The R7 is the only chip (I think) that's 90% of a 6900K and 90% of a 7700K for under $500.

Maybe the R5 1600X will be 80% of both for under $250. I think it'll be the only chip that would be in that category. And I think you'll find many more people willing to spend $250 than $500 (or $400 for the 1700X or $330 for the 1700).

As for me, I actually want the extra 2 cores so I'm willing to pay for a 1700 and OC it to 3.8-3.9 GHz for gaming. But for most, if the 1600X is as good as I think it will be, it'll be an absurdly good chip for the price.

 

[CHADBOGA]

This could be the chip I recommend a friend of mine get when he next upgrades his computer.

Would like to see some confirmation that all the motherboard gremlins have been sorted out first though.

 

[nathanddrews]

Wait, if the scheduling/CCX handling is what holds R7 back, wouldn't the same be true of R5?

 

[StinkyPinky]

To me the 1600 and 1600x will be the cpus many gamers actually buy. I'm not sure how far the 1600x will overclock though, as we have already seen the x model cpus seem to be factory overclocked to their near limit. I think the 1600 could be an awesome gaming cpu.

 

[unseenmorbidity]

quad core is 2+2, hmmm, I was hoping for a just a single ccx
interesting that one quad core is with 16MB and the other 8MB l3, looking at these specs I assumed the 8MB could be a single ccx, but by the article it sounds like they are all 2+2.

The 1400 might be a single ccx....

How stupid would that be? The 1400 might be the best r5 for gaming.

Why would amd use 2+2? Just to save a few bucks on silicon?

 

[Rngwn]

One thing though, the Wraith Spire that shipped with The 1600 appears to be without RGB. Hope this is not true.

 

[krumme]

Wait, if the scheduling/CCX handling is what holds R7 back, wouldn't the same be true of R5?

The ccx issue is way overblown. In bf1 an r7 is more or less the same as a 6900.
That is if you run the ram at 2933.
Look here
http://www.pcgameshardware.de/Battl...eld-1-They-Shall-not-Pass-Benchmarks-1223170/

I think a lot of the bad reviews about gaming is due to 2133 ram. Ofcource it makes the ccx issue bigger than it ought to be but also enhance the latency problem zen have.
Now there is games that is more or less single threadded and more of an ram benchmark than a cpu bm. Like far cry that needs some work. There is a few high profiles games that needs some work.

But aside from that what i think will happen is amd technical marketing will supply the bm kits with 2933 ram or perhaps even 3200 kits.
Now take a 6c 1600x and i think the reviews will be far more positive. Most games today just dont use all the threads.

When r7 was launched ram profiling was bad due to eg asus bios first beeing uploaded 28 feb. We have seen many 2133 reviews. They tend to be spammed here also.

So the end result might be some reviews saying this is a fantastic cpu while its just a cpu with 2 less cores. Granted if you dont oc thats a mighty fine offer. If you oc i think 80 usd for 2 cores extra in the 1700 is a steal.

I think with those low prices buying anything less than 6 cores 12t even for only gaming is outright stupid.

The 1600x will last many years as a gamer cpu. I think its a safe bet.

 

[Ancalagon44]

The 1400 might be a single ccx....

How stupid would that be? The 1400 might be the best r5 for gaming.

Why would amd use 2+2? Just to save a few bucks on silicon?

The Anandtech article says that AMD will only ever use 2 + 2, never 3 + 1 or 4 + 0.

Myself, I think the Ryzen 5 1500X might be the budget gaming champion. Relatively high clocks compared to other Ryzen products, plus 200Mhz XFR, plus budget selling price, plus 4C/8T. It should do pretty well for its price in games.

 

[unseenmorbidity]

The Anandtech article says that AMD will only ever use 2 + 2, never 3 + 1 or 4 + 0.

Myself, I think the Ryzen 5 1500X might be the budget gaming champion. Relatively high clocks compared to other Ryzen products, plus 200Mhz XFR, plus budget selling price, plus 4C/8T. It should do pretty well for its price in games.

The whole point of the design was 4 core modules that were scalable. Yet, they throw that away to save a few bucks.

Spend a dollar to save a dime.

Why not make the 1500x one ccx, and have the 1400 split? Are yields that bad?

 

[Valantar]

The whole point of the design was 4 core modules that were scalable. Yet, they throw that away to save a few bucks.

Spend a dollar to save a dime.

Why not make the 1500x one ccx, and have the 1400 split? Are yields that bad?

Simple answer: the only die AMD is producing at the moment has two CCXs. I'd bet fusing off a whole CCX is far more complex than fusing off individual cores (how does this affect the memory controller? Interconnects?).

Not to mention that the chance of a die with 3 or 4 defective cores on one CCX, and 4 perfect ones on the other seems ... small. Tiny, in fact. And any die with 3 functional cores per CCX can be sold as a 6/12 part, with higher margins.

We'll probably see a single-CCX die in time (though it might be limited to APUs, or APUs with fused-off GPUs like the Athlon X4 range), but I doubt we'll see any single-CCX implementations of dual-CCX dice.

 

[nathanddrews]

The ccx issue is way overblown. In bf1 an r7 is more or less the same as a 6900.
That is if you run the ram at 2933.
Look here
http://www.pcgameshardware.de/Battl...eld-1-They-Shall-not-Pass-Benchmarks-1223170/

I think a lot of the bad reviews about gaming is due to 2133 ram. Ofcource it makes the ccx issue bigger than it ought to be but also enhance the latency problem zen have.
Now there is games that is more or less single threadded and more of an ram benchmark than a cpu bm. Like far cry that needs some work. There is a few high profiles games that needs some work.

Overblown, yes, but still an issue considering it needs significant workarounds: buying faster/expensive RAM to reduce the impact and developer optimization to eliminate the impact.

That R5 1400 looks like a prime candidate for a tiny gaming build!

 

[krumme]

Overblown, yes, but still an issue considering it needs significant workarounds: buying faster/expensive RAM to reduce the impact and developer optimization to eliminate the impact.

That R5 1400 looks like a prime candidate for a tiny gaming build!

3000 ram is cheap these days the problem is memory profiling and uefi update is still ongoing work. In may we will know where we stand as amd will come with a new uefi then. If some standard 3000 or 3200 ram will then work at 2933 i think its a minor issue. But granted if the situation continues and you need expensive ram it certainly effects cost and is a drawback. If zen have a drawback right now its selecting and buying the right ram. It is very important.

As for optimizations its a new platform new core and cache typology so everyone will have to do a bit of work to get the last bit. Why on earth should it be different? Its not a skl copy. If anything the cpu seems fairly resistant to many if not all loads unlike all prior gen from k6 k7 k8 bd whatever.

 

[Agent-47]

The whole point of the design was 4 core modules that were scalable. Yet, they throw that away to save a few bucks.

Spend a dollar to save a dime.

Why not make the 1500x one ccx, and have the 1400 split? Are yields that bad?

these are defective CPUs. if there is a errata in L3, they make it into 4+0, if it's the core, then they go with 3+3 or 2+2 depending on the extent of the bugs. if they did not do so, they would be throwing away the chip. so by salvaging these chips, they are actually saving a low of money.

The larger the die, the probability of an error is higher. hence even intel does it with their BWE cores: those 6/8 cores are the same 10 core die as x6950.

 

[nathanddrews]

I do this when I make cookies. The best looking ones are placed out front for all to see, but the cookies everyone eats are a mix of underdone, overdone, and broken. They mostly taste the same.

 

[leoneazzurro]

these are defective CPUs. if there is a errata in L3, they make it into 4+0, if it's the core, then they go with 3+3 or 2+2 depending on the extent of the bugs. if they did not do so, they would be throwing away the chip. so by salvaging these chips, they are actually saving a low of money.

The larger the die, the probability of an error is higher. hence even intel does it with their BWE cores: those 6/8 cores are the same 10 core die as x6950.

Yes and no, in the sense that AMD is fabless now, so it depends on the type of agreement they have with GF. They can i.e. pay for good die and much less for downgrades and in that sense is probably GF that saves most of money (and AMD also pays less the lower binned chips, of course). The savings is all on AMD side if they pay for wafer, and not for good die but I think that is unlikely.

 

[unseenmorbidity]

Simple answer: the only die AMD is producing at the moment has two CCXs. I'd bet fusing off a whole CCX is far more complex than fusing off individual cores (how does this affect the memory controller? Interconnects?).

Not to mention that the chance of a die with 3 or 4 defective cores on one CCX, and 4 perfect ones on the other seems ... small. Tiny, in fact. And any die with 3 functional cores per CCX can be sold as a 6/12 part, with higher margins.

We'll probably see a single-CCX die in time (though it might be limited to APUs, or APUs with fused-off GPUs like the Athlon X4 range), but I doubt we'll see any single-CCX implementations of dual-CCX dice.

Yah, I am saying they should have done another run where they only made 4 cores, and they sold the bad 8 cores as 2+2 1400s.

Right now, it's quite possible the R3 will be better than 1500x for games.

 

[raghu78]

Ryzen 5 would benefit from the 1 month time AMD and motherboard partners have to improve higher memory speed support and stability. I think by now its very clear that Ryzen benefits hugely from DDR4 2933+ Mhz speeds as the inter CCX communication runs at half the memory speed. I think once Ryzen motherboards have stable support for DDR4 2933 / DDR4 3200 they would be much more impressive. We are seeing close to 10% performance improvement simply from moving from DDR4 2133 to 2933 Mhz alongwith HPET on and High performance Plan.

https://community.amd.com/community...4/tips-for-building-a-better-amd-ryzen-system

I think R5 1600x should perform very well in gaming against 7600k as more and more games are now capable of using 8+ threads. In fact even with such a massive clock speed lead and higher IPC 7700k is just on par with 1800x on avg across a range of games.

https://www.computerbase.de/2017-03/ryzen-windows-7-benchmark-core-parking/

AMD needs to work diligently with motherboard vendors to improve firmware support for high speed DDR4 memory.