Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

[french toast]

Definitely. The moment I saw 3.6/4.0 for 1600X, I realized that this is going to be my CPU.

Love at first sight

 

[bjt2]

If you use a lower Vcore, you sit at lower Fmax, that's fairly straightforward when your CPU sits at under 100 Kelvins.

I added the sentence "if the voltage is exaggerated". No one know the right voltage to apply to a 1800X under LN2, even the guys who set that record. They set an high voltage and an high clock and tryed to bench. You must test to know the minimum voltage. Maybe it runs at 5.2GHz even with 1.6V, so you can try 5.5GHz with 1.7V... Who knows.

 

[lolfail9001]

I added the sentence "if the voltage is exaggerated".

The only thing exaggerated voltage does in this case is speed up the death of CPU, nothing else.

Who knows.

These guys probably do, considering they did 10 attempts that evening alone and god knows how many in mobo makers HQ, most likely.

 

[beginner99]

Love at first sight

Even better if you could unlock cores...good old times.

 

[BeepBeep2]

I added the sentence "if the voltage is exaggerated". No one know the right voltage to apply to a 1800X under LN2, even the guys who set that record. They set an high voltage and an high clock and tryed to bench. You must test to know the minimum voltage. Maybe it runs at 5.2GHz even with 1.6V, so you can try 5.5GHz with 1.7V... Who knows.

As an overclocker, I know you are very wrong.

Overclockers 99.999% of the time test CPU for minimum voltage first, you are absolutely silly to think they just set a random high voltage and random clock speed, especially when so close to the 8-core CB15 record.

It is not hard to roughly test for minimum voltage either - set voltage (say, 1.5-1.6v) and keep pushing clockspeed, then increase voltage and clockspeed until a maximum speed is reached or the chip goes kaboom.

The leaked screen shows at least 12 runs with scores increasing.

 

[Insert_Nickname]

Its the latter. Rye-Zen. A Rye that's become totally Zen. I agree, even the "Risen" pronunciation would have deep philosophical meaning in the context of AMD's rise back to relevance/dominance.

Pure speculation, but the double meaning might be why they chose Ryzen in the first place. My own language has plenty of words with double (and even triple) meanings, some of which are further context sensitive.

 

[nathanddrews]

He is Ryzen indeed!

And to think just a month ago I almost spent $350 on a 5820K. I better get rid of that 2011-v3 board while it's still worth something.

 

[Dresdenboy]

Remember the old 1188 cinebench leak that was dismissed, well it was the very first ES 2.8ghz ACT

1537 * 2.8 / 3.7 = 1183 ~ 1188

Performance-wise yes. The ES string hinted at 3.15GHz base. But there are many possible reasons for it to work this "slow", like: no turbo, SMT and/or uop$ disabled (bug workaround in BIOS).

BTW, the OP of the locked thread likes AMD's new processor. So he didn't update to make sure to get his preordered part.

 

[007ELmO]

I think a lot of us are sitting and waiting/watching this thread and others. Just too much information!

Maybe I liken it to my switch from TWC to WOW cable that's about to happen. TWC is charging me $60 a month for 30Mbps. WOW will charge me $70 a month for 500Mbps.

 

[bjt2]

As an overclocker, I know you are very wrong.

Overclockers 99.999% of the time test CPU for minimum voltage first, you are absolutely silly to think they just set a random high voltage and random clock speed, especially when so close to the 8-core CB15 record.

It is not hard to roughly test for minimum voltage either - set voltage (say, 1.5-1.6v) and keep pushing clockspeed, then increase voltage and clockspeed until a maximum speed is reached or the chip goes kaboom.

The leaked screen shows at least 12 runs with scores increasing.

I didn't have that information. If you say that is at least the 12th try, then you are right.
Where are those leaked screen?

 

[DrMrLordX]

Who buys PC's anymore...

Me, yesterday.

AMD needs to attack the server market with these chips.

Duh. We're the guinea pigs for that.

 

[BeepBeep2]

I didn't have that information. If you say that is at least the 12th try, then you are right.
Where are those leaked screen?

Sorry for the bad resolution, this is a 360p reupload of the Youtube video which had the scores.

Also, it looks like Sami Mäkinen "Macci" (legendary extreme overclocker, AMD employee) was there and I'm pretty sure he would know what he is doing.
Mark Papermaster was also standing right there I believe, and AMD had internally broken 2K Cinebench a couple months ago.

John Sandstrom (ASUS Engineer) is the one with his arm up.

 

[malitze]

German magazine computerbase did an interesting review about CPU core scaling in modern games in preparation for their Ryzen coverage. They come to the conclusion that modern titles do profit from more cores very well contrary to common believe.

 

[Douglas Silva]

Where are the games benchmark? Lol

 

[leoneazzurro]

In the very same page, you have to click "nachstes" just above the last graph to scroll among the various tests

 

[Mercennarius]

German magazine computerbase did an interesting review about CPU core scaling in modern games in preparation for their Ryzen coverage. They come to the conclusion that modern titles do profit from more cores very well contrary to common believe.

I think at this point it is common knowledge that many newer games, especially DX12 or Vulkan titles, benefit with more cores/threads.

 

[MrTeal]

And here i am thinking that the real sweetspot is r7 1700. Though in hindsight it is like full 100 bucks for 2 more cores.

Heh, full $100 for two more cores.
Coming from 5820K to 5960X and 6800K to 6900K to 6950X, that seems like an absolute bargain.

 

[Blitzvogel]

I was hoping for some info on Ryzen 3 and 5 yesterday, but it was all about beating Intel, hence going big. Ryzen 3, especially the version with SMT and at a good price could finally put dual cores to rest. I wonder if Ryzen 3 will be die harvested Ryzen 7s (a la FX-4100s) or if it'll be it's own die.

 

[Andrei.]

German magazine computerbase did an interesting review about CPU core scaling in modern games in preparation for their Ryzen coverage. They come to the conclusion that modern titles do profit from more cores very well contrary to common believe.

Fantastic article. Demolishes the claim that single-threaded performance still trumps gaming. Ryzen will clean the floor with Intel in all segments, gaming included.

 

[.vodka]

I think at this point it is common knowledge that many newer games, especially DX12 or Vulkan titles, benefit with more cores/threads.

For many of us it is, but there's still LOTS of people out there with the outdated 2010 mentality that 4C4T (i5) is all you need for gaming. They keep spreading that "advice" everywhere.

Couldn't be any further from the truth.

Buying a quad core in any shape or form in 2017 for demanding tasks is squandering money. Quad cores are the new dual cores.

 

[Arachnotronic]

For many of us it is, but there's still LOTS of people out there with the outdated 2010 mentality that 4C4T (i5) is all you need for gaming. They keep spreading that "advice" everywhere.

Couldn't be any further from the truth.

Buying a quad core in any shape or form in 2017 for demanding tasks is squandering money. Quad cores are the new dual cores.

7700K is fine, 4 core/8 thread at high clock. But 7600K value prop really looks iffy with Ryzen around at similar price points.

 

[LTC8K6]

For many of us it is, but there's still LOTS of people out there with the outdated 2010 mentality that 4C4T (i5) is all you need for gaming. They keep spreading that "advice" everywhere.

Couldn't be any further from the truth.

Buying a quad core in any shape or form in 2017 for demanding tasks is squandering money. Quad cores are the new dual cores.

Including RyZen quad cores? A lot of people don't game or do anything more taxing than occasional video or photo work.

 

[GroundZero7]

And here i am thinking that the real sweetspot is r7 1700. Though in hindsight it is like full 100 bucks for 2 more cores.

That depends on what the 6 core 1600X will OC to with a Corsair H115i. If it can hit +200mhz it might be worth it for the increased IPC. We don't know the thermal characteristics of these chips.

If the transistor density(heat) is overwhelming the soldered lid...4.8B transistors..holy fi*g s*it!!. Kaby lake has only 1.75B or so. What if Ryzen is thermally constrained not by transistor density but by the amount of heat that can pass through the soldered lid? Rzn 6 core might OC very well if that's the case.

You start upping the voltage on 4.8 billion transistors and you are going to hit a ceiling very fast. The 6 core will have less active transistors, and the 4 core even more so.

Just my ignorant 2c. I hope someone far more knowledgeable than me could expand on this.

 

[BeepBeep2]

What if Ryzen is thermally constrained not by transistor density but by the amount of heat that can pass through the soldered lid? Rzn 6 core might OC very well if that's the case.

Considering LN2 is currently hitting 5.2 GHz, probably not. It may overclock higher because of less heat output, but probably not more than 100-200 MHz.

 

[coercitiv]

7700K is fine, 4 core/8 thread at high clock. But 7600K value prop really looks iffy with Ryzen around at similar price points.

Amazing how this idea keeps coming up: if 4c/4t i5 is no longer enough for today, then i7 4c/8t is definitely not enough for tomorrow. Either we accept some people can find proper value in 4c/4t chips priced $100 less, or we stop deluding ourselves that 7700K is a CPU that will last enough to be worth the premium. If i5 reached a performance threshold where it starts to lag behind, it's only a matter of time the current mainstream i7 will follow.