Official AMD "Ryzen" CPU Discussion

[TSO]

Yeah, ultimately pricing always determines whether or not a component is a good buy, but I'm sure certain AMD fans were hoping they would be more competitive than that on an absolute scale. I think some were hoping that with DX12 advantages it might rival the GTX 1080 Ti.

Redgamingtech is saying they're releasing three levels of cards and the three of them will beat the 1070, 1080, and 1080 ti respectively. But he's far from unbiased.

 

[Slobodan]

Yeah, ultimately pricing always determines whether or not a component is a good buy, but I'm sure AMD fans were hoping they would be more competitive than that on an absolute scale. I think some were hoping that with DX12 advantages it might rival the GTX 1080 Ti.

Redgamingtech is saying they're releasing three levels of cards and the three of them will beat the 1070, 1080, and 1080 ti respectively. But he's far from unbiased.

I have seen mentions of "big Vega" and "little Vega" before, with the latter possibly being the one shown in that article, it would be incredibly disappointing that after all this time, the Vega barely beats the 1070.

I hope there are different tiers.

 

[Madmick]

I have seen mentions of "big Vega" and "little Vega" before, with the latter possibly being the one shown in that article, it would be incredibly disappointing that after all this time, the Vega barely beats the 1070.

I hope there are different tiers.

That's weird. When I hear those terms I think of ARM processors: "big" and "little" configurations of CPU clusters. It's all the rage. They pair a few cores with high GHz and high performance with another few cores that are low GHz, so they generate less heat and use less energy, and only kick in when the first two cores can't handle a load.

Maybe they did something like that with Vega. I haven't read much about Vega. I don't like committing rumor to memory. It makes remembering what's real harder. That's why I forget tidbits like Kaby Lake "K" processors shipping without stock coolers for the first time in many generations, or that the R5-1400 has half the L3 cache as the R5-1500X (and so isn't a better value even for overclockers), or that the GTX 1060 3GB has a different core config than the 1060 6GB altogether as I did in this thread. It's hard enough remembering what's real.

 

[Slobodan]

That's weird. When I hear those terms I think of ARM processors: "big" and "little" configurations of CPU clusters. It's all the rage. They pair a few cores with high GHz and high performance with another few cores that are low GHz, so they generate less heat and use less energy, and only kick in when the first two cores can't handle a load.

Maybe they did something like that with Vega. I haven't read much about Vega. I don't like committing rumor to memory. It makes remembering what's real harder. That's why I forget tidbits like Kaby Lake "K" processors shipping without stock coolers for the first time in many generations, or that the R5-1400 has half the L3 cache as the R5-1500X (and so isn't a better value even for overclockers), or that the GTX 1060 3GB has a different core config than the 1060 6GB altogether as I did in this thread. It's hard enough remembering what's real.

Yeah that's what I think of as well when I hear big/little lol

I know they showed Vega running DOOM 4K at 60+fps and that was with the Fiji drivers I believe, I believe that is the "big" Vega, the upper tier one. I dunno, I'm just hoping that's some early test sample with un-optimised drivers or something.

 

[GhostZ06]

We talking about the Vega 490 and 490x?



well wait, which one is the Fury 2?

 

[jefferz]

Yeah that's what I think of as well when I hear big/little lol

I know they showed Vega running DOOM 4K at 60+fps and that was with the Fiji drivers I believe, I believe that is the "big" Vega, the upper tier one. I dunno, I'm just hoping that's some early test sample with un-optimised drivers or something.

AMD does some shady crap to inflate numbers when showing off hardware before launches. I wouldn't rely on their numbers.

 

[Slobodan]

AMD does some shady crap to inflate numbers when showing off hardware before launches. I wouldn't rely on their numbers.

What's shady about it?

 

[Slobodan]

We talking about the Vega 490 and 490x?



well wait, which one is the Fury 2?

I dunno, it's all confusing now.

 

[TSO]

AMD does some shady crap to inflate numbers when showing off hardware before launches. I wouldn't rely on their numbers.

My only experiences with them have been the opposite. They say something will do X, but you buy it and it does X2.

 

[Madmick]

We talking about the Vega 490 and 490x?

well wait, which one is the Fury 2?

That's what we don't know. Usually early leaked figured for a new microarchitecture are for the flagship release. For NVIDIA, they always that splash with the "80" card: GTX 1080, GTX 980, GTX 780, GTX 680, GTX 580, etc. Ultimately, that won't be the fastest release on the microarchitecture, since they'll refine it with the 1080 Ti and Titan releases, but that's the usually the reference for the microarchitecture as a whole when it is debuted.

I think most of assuming that about this Vega card, and this will be the first true generational leap in terms of technology.

My only experiences with them have been the opposite. They say something will do X, but you buy it and it does X2.

Oh, I think he's just talking about the marketing circus AMD does their best to invent around any new release. It's irritating when they hype CPUs like the FX-9590 that was literally little more than a pre-overclocked FX-8350. They've done that shit more than once. He's probably thinking of stunts like the AMD employee accidentally "exposing" a screen with OC results that just have no relevance for the mainstream market even if it wasn't a bullshit score that only maxed a single thread (because the mainstream isn't using de-lidded processors on nitrogen-cooled test benches for PC gaming). So there they are trying to get people talking about 5GHz overclocking due to raw potential, and since that time, the community appears to have determined AMD themselves built in some sort of voltage governor that prevents Ryzen from overclocking beyond the 4.1 GHz wall in the real world.

Of course, with Ryzen itself, that criticism against AMD isn't as compelling, and if Intel actually had a rabbit in their hat to hype, they would have done the same thing. It's wouldn't be wise for them to stir up market hype around Kaby Lake. How would that go? I'm reminded of the scene with Christopher Guest's rocker character talking about how Spinal Tap's shows are the loudest in the land because their custom speakers' volume dials "go to 11". That pretty much would have been Intel's marketing campaign for Kaby Lake. "And check out our new processors...they have a 7 in front of them! Brand new shit!!"

 

[TSO]

That's what we don't know. Usually early leaked figured for a new microarchitecture are for the flagship release. For NVIDIA, they always that splash with the "80" card: GTX 1080, GTX 980, GTX 780, GTX 680, GTX 580, etc. Ultimately, that won't be the fastest release on the microarchitecture, since they'll refine it with the 1080 Ti and Titan releases, but that's the usually the reference for the microarchitecture as a whole when it is debuted.

I think most of assuming that about this Vega card, and this will be the first true generational leap in terms of technology.

Oh, I think he's just talking about the marketing circus AMD does their best to invent around any new release. It's irritating when they hype CPUs like the FX-9590 that was literally little more than a pre-overclocked FX-8350. They've done that shit more than once. He's probably thinking of stunts like the AMD employee accidentally "exposing" a screen with OC results that just have no relevance for the mainstream market even if it wasn't a bullshit score that only maxed a single thread (because the mainstream isn't using de-lidded processors on nitrogen-cooled test benches for PC gaming). So there they are trying to get people talking about 5GHz overclocking due to raw potential, and since that time, the community appears to have determined AMD themselves built in some sort of voltage governor that prevents Ryzen from overclocking beyond the 4.1 GHz wall in the real world.

Of course, with Ryzen itself, that criticism against AMD isn't as compelling, and if Intel actually had a rabbit in their hat to hype, they would have done the same thing. It's wouldn't be wise for them to stir up market hype around Kaby Lake. How would that go? I'm reminded of the scene with Christopher Guest's rocker character talking about how Spinal Tap's shows are the loudest in the land because their custom speakers' volume dials "go to 11". That pretty much would have been Intel's marketing campaign for Kaby Lake. "And check out our new processors...they have a 7 in front of them! Brand new shit!!"

My 1700 runs at 4.212 stable. It gets hot though so I usually keep it at 3.8. That's on a CPU they advertised as 3.0 that could OC to 3.4, but it came already running 3.4 and can boost to 4.2. To me, I got double what I paid for.

 

[Slobodan]

My 1700 runs at 4.212 stable. It gets hot though so I usually keep it at 3.8. That's on a CPU they advertised as 3.0 that could OC to 3.4, but it came already running 3.4 and can boost to 4.2. To me, I got double what I paid for.

What cooler do you have?

 

[TSO]

What cooler do you have?

Stock wraith spire.

 

[Slobodan]

Stock wraith spire.

Niiice



I've been wanting to try overclocking but wasn't sure if I should go and buy an aftermarket cooler or just keep the stock Wraith Spire...

You have a B350 board right? What are your voltages?

 

[TSO]

Niiice



I've been wanting to try overclocking but wasn't sure if I should go and buy an aftermarket cooler or just keep the stock Wraith Spire...

You have a B350 board right? What are your voltages?

The BIOS on my b350 ASUS prime is weird, you can't set exact voltage, you set a range and I just set it to the widest range available.

 

[Madmick]

My 1700 runs at 4.212 stable. It gets hot though so I usually keep it at 3.8. That's on a CPU they advertised as 3.0 that could OC to 3.4, but it came already running 3.4 and can boost to 4.2. To me, I got double what I paid for.

That's groovy, and it sounds like you got a real bin winner, but I'm not imagining the voltage throttling, TSO. I've literally watched guys like Jay hit it in their videos. If you had bought a better CPU cooler, you would have just been wasting your money, dude. You hit the enthusiast performance:value sweet spot with your CPU purchase. Couldn't have been more perfect.
https://www.reddit.com/r/Amd/comments/5xk2bz/39_41_volt_wall_on_1700_1700x_and_1800x/


Screw the second link for the Reddit survey even though it confirms this. Fuck self-reporting. Unreliable.

More reliable: benchmarks. There is yet to be turned in an OC on UserBenchmark with a stable clock for the R7-1700 above a 4.05 GHz base clock. Check out the top scores for Ryzen 7 of all their benchmarks on UserBenchmark. These are confirmed clocks; able to stably complete the benchmark. Scroll down the top few hundred descending from the best results. The top clocked processor doesn't always notch the best score, but you see page after page after page of highest cores on the board at 4.0 GHz for all three processors. Meanwhile, across more than ten thousand Ryzen 7 benchmarks, there are fewer than a dozen R7-1800X's that were able to complete the benchmark at a frequency of 4.15GHz, and only one at 4.2GHz. None of the 1700/1700X systems could break the 4.1GHz wall:

R7-1700 = 4.05GHz [Top MC score = 4.05 GHz]
http://www.userbenchmark.com/UserRun/3554540
http://cpu.userbenchmark.com/AMD-Ryzen-7-1700/Rating/3917

R7-1700X = 4.05 GHz [Top MC score = 3.85 GHz]
http://www.userbenchmark.com/UserRun/3463396
http://cpu.userbenchmark.com/AMD-Ryzen-7-1700X/Rating/3915

R7-1800X = 4.2 GHz [Top MC score = 4.00 GHz]
http://www.userbenchmark.com/UserRun/3490818
http://cpu.userbenchmark.com/AMD-Ryzen-7-1800X/Rating/3916

 

[TSO]

That's groovy, and it sounds like you got a real bin winner, but I'm not imagining the voltage throttling, TSO. I've literally watched guys like Jay hit it in their videos. If you had bought a better CPU cooler, you would have just been wasting your money, dude. You hit the enthusiast performance:value sweet spot with your CPU purchase. Couldn't have been more perfect.
https://www.reddit.com/r/Amd/comments/5xk2bz/39_41_volt_wall_on_1700_1700x_and_1800x/
View attachment 224973

Screw the second link for the Reddit survey even though it confirms this. Fuck self-reporting. Unreliable.

More reliable: benchmarks. There is yet to be turned in an OC on UserBenchmark with a stable clock for the R7-1700 above a 4.05 GHz base clock. Check out the top scores for Ryzen 7 of all their benchmarks on UserBenchmark. These are confirmed clocks; able to stably complete the benchmark. Scroll down the top few hundred descending from the best results. The top clocked processor doesn't always notch the best score, but you see page after page after page of highest cores on the board at 4.0 GHz for all three processors. Meanwhile, across more than ten thousand Ryzen 7 benchmarks, there are fewer than a dozen R7-1800X's that were able to complete the benchmark at a frequency of 4.15GHz, and only one at 4.2GHz. None of the 1700/1700X systems could break the 4.1GHz wall:

R7-1700 = 4.05GHz [Top MC score = 4.05 GHz]
http://www.userbenchmark.com/UserRun/3554540
http://cpu.userbenchmark.com/AMD-Ryzen-7-1700/Rating/3917

R7-1700X = 4.05 GHz [Top MC score = 3.85 GHz]
http://www.userbenchmark.com/UserRun/3463396
http://cpu.userbenchmark.com/AMD-Ryzen-7-1700X/Rating/3915

R7-1800X = 4.2 GHz [Top MC score = 4.00 GHz]
http://www.userbenchmark.com/UserRun/3490818
http://cpu.userbenchmark.com/AMD-Ryzen-7-1800X/Rating/3916

I ran my bench at 3.8 but I guess I can go for a 4.212 to see if it flies.

 

[TSO]

Well, that didn't go so well. It crashed while running the test at 4.212, and is now rebooting.

 

[PEB]

AMD already getting ready to show their next generation high end chip. Get this it has 4096 pin socket and 16 cores. But the real winner is it has 2 times as much level 2 cache as the rizen 1800/1700 series.

They are going to be shooting for 12, 16, 32 cores in the high end enthusiast market. They are also targeting it for server market. But here is the part that kind bothers me the enthusiast chip is not pin compatible with the 4096 pin server chip.

AMD of course stated that there are features in the server chip that is not needed on the high end enthusiast chip.

It's going to have quad DDR4 channels

WCCFTECH Whitehaven
(Threadripper) Summit Ridge
(Ryzen)
Cores Up to 16 Up from 8,Threads Up to 32 Up from 16, Base Clock 3.1GHz 3.6GHz ,Boost Clock 3.6GHz 4.0GHz L3 Cache 32MB up from 16MB,TDP Up To 180W Up from 95W, DDR4 Channels Quad Dual
Socket S3 (LGA) VS AM4 (PGA)
Launch Mid 2017 Q1 2017

http://wccftech.com/amd-ryzen-16-core-threadripper-whitehaven-4094-socket/

In other news Intel reworking their lines to continue to lead AMD. At least improvements on their CPU lines.

"
Intel Skylake X Core i9-7920X, Core i9-7900X, Core i9-7820X, Core i9-7800X Mega-Tasking CPUs Leaked – Kaby Lake X Core i7-7740K and Core i7-7640K To Be Entry Level LGA 2066 HEDT Chips"

http://wccftech.com/intel-skylake-x-core-i9-7920x-7900x-7820x-7800x-x299-leaked/

jefferz one of those god awful Acer monitors LOL.

"Starting with the flagship, we have the Core i9-7920X processor. This chip is a juggernaut featuring a total of 12 cores and 24 threads. The total cache on this behemoth is 16.5 MB (L3). Although lower than whole cache featured on previous HEDT processors, the new cache runs more efficiently and reduces chip size and cost while delivering better performance. The chip will ship with a TDP of 160W."

 

[jefferz]

The chip will ship with a TDP of 160W."

Holy shit. Time to dust off the old bulldozer cpu cooler.