Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

[CentroX]

stock of 1700 is 3ghz, so i dont see any wrong with it being that behind.

 

[lolfail9001]

1700x vs 7700k

1700x is winning by a mile!

Look at this CPU-Z and tell me with straight face it is real.

 

[richierich1212]

A 1700 that isn't even OC'd!? lol Get real...

1700x vs 7700k

1700x is winning by a mile!

Damn that's a nice result in a popular game. And can be better if they were able to OC a little.

 

[.vodka]

EDIT: Probably fake, actually.

Probably. Could you post a link to the actual video instead of a screenshot? Seems fabricated, as this other video posted by unseenmorbidity has Ryzen well ahead of 7700k by over 20 FPS in actual gameplay. Still, it's done on a qualification sample, not retail silicon.

Although CPUz looks funny in this video...

edit: oh, misread. It's the R7 1700, I thought I read an X in there. Should get glasses.

It makes sense for it to post those numbers, it's 3GHz base

 

[Glo.]

The core clock is 3.6 on that QS, so it is 1800X, actually.

 

[looncraz]

So Ryzen at about 9.4/GHz. The data I have show 4790 @ 10.4/GHz in 64-bit and 10.2/GHz in 32-bit. Both with gcc 5.2 and -O3 (IIRC). Not exactly the same compiler revision and not the same flag, but I don't expect more than 10% difference.

I hope someone will make a fair Intel vs AMD comparison with the same compiler and the same set of flags!

I have a few apps that showed more than a 10% difference when I switched to GCC5 from GCC4. And a few that will benefit nearly as much from going to -O3 from -O2.

Most, naturally, exhibit only minor differences (either good or bad).

 

[lolfail9001]

Could you post a link to the actual video instead of a screenshot

https://youtu.be/3mJIP0-1xDo

Do you recognize the mobo?

 

[CentroX]

 

[lolfail9001]

What is it based on? Speculation on userbench with unknown clocks?

Still, it's done on a qualification sample, not retail silicon.

Also, QS is retail silicon bar naming.

 

[.vodka]

Source?

Please post source or where you got it from every time you find a benchmark like this.

https://youtu.be/3mJIP0-1xDo

Do you recognize the mobo?

He says he's using the Crosshair VI Hero. It's next to its box

Also, QS is retail silicon bar naming.

Right. Excellent then

 

[lolfail9001]

He says he's using the Crosshair VI Hero. It's next to its box

The point being is that it does have pretense of having proper AM4 mobo and proper AM4 chip in it. Pretense, not evidence.

It makes sense for it to post those numbers, it's 3GHz base

It looks like QS 1700X, though, not 1700. I did not listen to video whether they downclocked it though.

 

[BeepBeep2]

To have linearity in score, all clocks must scale linearly: CPU, RAM and NB. This is the case, probabily, with the record.
Any clock not increased causes underlinear behaviour.
They already performed the best case. In reality, raising only CPU clock gives diminished returns, namely you need even higher CPU clock to score a certain value.

Not sure about Ryzen, but for example past AMD CPUs such as Phenom II had CPUNB set to 2 GHz when 3 GHz (50% increase) was possible on air, close to 5 ghz (150%) possible on LN2. Of course, the CPU clock did not scale that well, so I think greater than linear scaling compared to AMD's baseline is possible.

If the DRAM multiplier was 24x or 26x here (DDR4-3400 or 3700) as well with drastically improved bandwidth / latency (overclockers are hitting DDR4-4000 CL10 on Intel platforms vs. DDR4-2400 CL15) that would also be a higher increase proportional to the CPU clock increase.

 

[looncraz]

I'm comparing the R7 1800x with the R7 1800x though. The only way that result can be explained, using simple maths, is that Ryzen SMT improves as you increase the clock; i.e. SMT gets better the more you OC it.

I don't know whether that is a viable conclusion.

I know that with X4 845 (Excavator) the CPU will continue to show ~3.8GHz even when the bus is being stretched 5% or more.

I think 1800X, with the cooling used, may have actually stretched the clock. That is basically like running at 3.4Ghz without being able to realize it.

I have a couple old AMD stock coolers which I am going to use to test the 1800X before attaching my waterblock. I wanted to represent as close to a worst-case situation as possible (a 95W TDP heatsink should do ) before representing high-end air (D15) and high-end water (EK Supremacy Evo, 3x140mm XSPC radiator, MCP35X pump, CPU-only loop).

 

[zinfamous]

"There are many fine justifications for more cores but Google Chrome"

As someone who can have 100 tabs open at one time.. you'd be surprised how sluggish google chrome can become on weak systems. Many sites can be javascript heavy and a bunch of intensive tabs open can eat up a good amount of resources.

my crappy 8gb i5whatsit Windows 8 machine at work is constantly running out of memory for Chrome. I have to keep 3 windows upon (because of various email accounts), and usually about 20 or so tabs across all the windows. ...and I don't think that is a lot of tabs.

 

[CentroX]

Source?

Please post source or where you got it from every time you find a benchmark like this.




He says he's using the Crosshair VI Hero. It's next to its box



Right. Excellent then

http://www.sweclockers.com/forum/post/16680954

 

[AtenRa]

But you are suggesting it throttles below base clock. Otherwise your input is irrelevant.

Nope, when i said its throttling down i mean it doesnt keep Turbo clocks. But it also doesnt keep running at 3.6GHz constant for the duration of the benchmark.

 

[.vodka]

http://www.sweclockers.com/forum/post/16680954

With the source being

http://wccftech.com/amd-ryzen-1700x-benchmarks-leaked-beats-kaby-lake-ipc/

Meh. "Userbench" from one of those crappy comparison sites. Downloaded the benchmark, put it through virustotal and it shows some detections.

 

[CatMerc]

Cinebench versions were different, so you can calm down for now.

Anyways, looks like Zen is actually VRM limited O_O. Or it has voltage/freq so extreme that VRMs give up.

I actually laughed when I saw Gibbo's results. To actually be limited by VRM's... That wall is amazing.

Shall we call it... Trump's Curve?

 

[CentroX]

what does VRM stand for anyway?

 

[GroundZero7]

what does VRM stand for anyway?

Voltage Regulator Module

my crappy 8gb i5whatsit Windows 8 machine at work is constantly running out of memory for Chrome. I have to keep 3 windows upon (because of various email accounts), and usually about 20 or so tabs across all the windows. ...and I don't think that is a lot of tabs.

I made the switch to Palemoon. Couldn't be happier. I hate Chrome.

 

[looncraz]

what does VRM stand for anyway?

Voltage Regulator Module.

 

[CentroX]

Voltage Regulator Module.

ok, and the asus prime pro has crappy VRM?

 

[DrMrLordX]

So against some older competition and Broadwell-E the 7700K was fastest in 2 games out of 14 and all were at stock clockspeeds (which should be a bonus on 7700K side). This pretty much tells us that lower clockspeed and more cores is the way to go even today.

Any idea how much the massive L3 cache helps Broadwell-E in games?

Ah, good of you to mention the L3. L3 already seems to make a big difference with the 5960x, so I'm assuming it does the same for Broadwell-E. It would be nice if reviewers could disable a few cores (but not the L3) to show us just how much a 6950x going from (for example) 10->6->4 cores (SMT active) affects gaming results, minimums, etc.

That doesnt change the fact that this segment is low volume and will not increase AMDs x86 market share vs Intel.

But it will have at least a positive impact on the Q1 and more on the Q2 (R5 release as well) 2017 financial results.

Right now it's as much about mind share as it is anything else. Everyone is infected by Ryzen fever. An octocore + SMT with Haswell/Broadwell IPC and clocks up to 4.2+ GHz for $499? Unbelievable! Yet they did it.

ok, and the asus prime pro has crappy VRM?

Summit Ridge will redefine everyone's concept of what is "crappy VRM".

 

[bjt2]

A 1700 that isn't even OC'd!? lol Get real...

1700x vs 7700k

1700x is winning by a mile!

The screen at the start says 6 core for ryzen, so it is not a 1700X, but still impressive.

 

[AtenRa]

Right now it's as much about mind share as it is anything else. Everyone is infected by Ryzen fever. An octocore + SMT with Haswell/Broadwell IPC and clocks up to 4.2+ GHz for $499? Unbelievable! Yet they did it.
.

Agreed, they entered a "new" market ($300 and up) that Intel has 100% of market share the last 6-7 years. They had to make a big splash, they had to disrupt the market