Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

[Hitman928]

NOTE: If you want something included in the OP, send me a pm with the info. I don't have time to monitor every post that is made. You can still post it in the thread, just note in the thread post that you have sent the pm to me to include it so people don't spam me if I don't add it right away. Thank you.



Performance

*credit computerbase.de

Power Consumption

*credit computerbase.de

Purchase

Newegg
Amazon

Reviews

Anandtech
Guru3d
HardOCP
Overclockers.com
PCPer
Sweclockers
Computerbase
Joker Productions (youtube)

 

[Hitman928]

Motherboards

Platform specs.

 

[sirmo]

Can we lock the other thread?

 

[parvadomus]

Every new leak make it look even better.
But.. I want a 4c/8t APU, with at least a gtx750 level integrated GPU, that would be insane.

 

[Absolute0]

My primary unanswered question is this... We have seen benchmarks from primarily 1600X, 1700X, and 1800X, which as I understand it, all have XFR and turbo up in speed according to the thermal headroom available. So, are the benchmarks we see performed at the nominal speeds of these ----X series processors, or at some unknown turbo speed, possibly 4.1, 4.2, 4.3 Ghz?

I'm sure someone must have compared the numbers on earlier Ryzen leaks (fixed clockspeed, ~3.3 Ghz) to the current leaks to extrapolate if these X-series chips are running close to the 3.8, 4.0 they are listed at or turboing higher?

> Obviously if they are running at their rated specs, that bodes well for performance.

> If they are turboing but only very minimally, then again it indicates performance is strong but perhaps there is not much overclocking headroom (or the XFR feature does not utilize it effectively).

> Lastly, if they are turboing substantially, that means that this high level of performance has been obtained by overclocking these chips probably close to the max OC supported by air. Which is good and bad, because (-) the competition might be overclocked to have a more fair comparison, but at least (+) there is significant OCing headroom.

 

[inf64]

My primary unanswered question is this... We have seen benchmarks from primarily 1600X, 1700X, and 1800X, which as I understand it, all have XFR and turbo up in speed according to the thermal headroom available. So, are the benchmarks we see performed at the nominal speeds of these ----X series processors, or at some unknown turbo speed, possibly 4.1, 4.2, 4.3 Ghz?

I'm sure someone must have compared the numbers on earlier Ryzen leaks (fixed clockspeed, ~3.3 Ghz) to the current leaks to extrapolate if these X-series chips are running close to the 3.8, 4.0 they are listed at or turboing higher?

> Obviously if they are running at their rated specs, that bodes well for performance.

> If they are turboing but only very minimally, then again it indicates performance is strong but perhaps there is not much overclocking headroom (or the XFR feature does not utilize it effectively).

> Lastly, if they are turboing substantially, that means that this high level of performance has been obtained by overclocking these chips probably close to the max OC supported by air. Which is good and bad, because (-) the competition might be overclocked to have a more fair comparison, but at least (+) there is significant OCing headroom.

XFR is not going to boost the all core load workloads.
We have seen 6C and 8C benchmarks and they perfectly match with all core Turbo clocks for each model : 3.4Ghz MT CB run for 6C ES and ~3.5Ghz 8C MT CB run for 1700x. As for XFR it only adds 100Mhz on top of ST Turbo which means 4.1Ghz for 1800x when ST benchmark is run and 3.9Ghz for 1700x when ST benchmark is run. This is confirmed via AMD's slide and from leaked 1700X R15 ST benchmark result. AMD slide states 162 @ max clock in ST while 1700X scored 154 @ XFR bost clock of 3.9Ghz. 154*4.1/3.9=162. XFR ST boost clock for 1800X is 4.1Ghz.

All core Turbo also seems to be one bin (+100Mhz) above base clocks of each model: ~3.7Ghz for 1800X and 3.5Ghz for 1700X. That was confirmed by the numbers from official leaked AMD slide and the previous leak for 1700X. AMD slides states 1601pts for 1800X while 1700X(3.4Ghz base, 3.5Ghz ACT) scored 1537. If you adjust the clock to 3.7ghz which is supposed ACT for 1800X you will get: 1537*3.7/3.5=1624pts which is ~1.4% deviation from AMD's result and can be easily attributed to other stuff like memory clock and timings.

 

[Hitman928]

snip

Great question, obviously we don't know for sure yet, but the estimates I've seen online calculated a very minor (100 MHz or so) bump from XFR on the 1700x for a single core. I don't think anyone without the chip knows what they run at with all cores running yet.

 

[Maverick177]

Any leaks about the single core OC apps?

 

[IEC]

If we're limited on overclocking more by the process/silicon than by cooling, I'll hold off on my water cooling purchases for now and wait to see OC results.

If we're talking a negligible difference between high end air (e.g. Noctua D15) vs AIO or custom water I'll skip the headache and go air.

 

[Doom2pro]

Can we lock the other thread?

It's locked now.

 

[Ryzenenthusiast]

Biased thread get locked. GJ Moderators...

 

[unseenmorbidity]

Biased thread get locked. GJ Moderators...

Only took like half a year...lol

 

[SPBHM]

that 162 single thread score is impressive for AMD, their previous best, the 5GHz 220W thing scored 110 points
also, thinking about support for fast DDR4, that's a thing that could actually make it faster than Haswell for gaming and so on (still, I'm not sure, maybe the caches are not as good or something deeper; need to wait for proper tests for that), also I'm worried about motherboards and overclock (with normal cooling), but so far I'm really impressed; Bulldozer was clearly terrible even before launch, this is looking completely different, a return to form, this is looking better than Phenom II was in 2009.

even the cheaper 4 core ones are looking interesting for a gaming/average use machine, you don't really need 8c in 2017 for that, but you can certainly use the very decent ST performance it seems to offer.

 

[.vodka]

If we're limited on overclocking more by the process/silicon than by cooling, I'll hold off on my water cooling purchases for now and wait to see OC results.

If we're talking a negligible difference between high end air (e.g. Noctua D15) vs AIO or custom water I'll skip the headache and go air.

Me too. If Ryzen absolutely requires water cooling to push it.. well, we'll see. If not I already have a decent air cooler (TRUE Spirit 140m Power, 6x8mm heatpipes) that I'd have to mod the backplate and bracket (it's just drilling 8 holes) to make it compatible with AM4. Amusingly after doing measurements and some planning I came across the updated Thermalright backplate and those extra holes by the old AM3 mount are exactly where I measured mine on the old backplate + bracket


I'd love to see reviews of a variety of boards to see how far each can push eight core Ryzen. I'd also love to see reviews of the lower end B350 boards to see how they fare when overclocking the budget 4 core and mid range six core models.


I mean we could be back to the golden era of building a powerhouse of a PC with a $70 B350 motherboard and a 4C8T OC'd Ryzen for cheap!

 

[Doom2pro]

Ryzen 8c/16t 3.4Ghz base Validation Sample (ZD3406BAM88F4) on bench in ASRock Motherboard (X370 Gaming Fatal1ty) w/BIOS menu.

0:21 Sample Image
1:25 Heatsink going on
1:55 BIOS Menu

English Summary:

"I am holding a Ryzen, it is working. Look at BIOS. We are under NDA so we can not provide more information. But stay tuned. We will provide more information as time goes by."

 

[cytg111]

Biased thread? What was wrong with the old one? Not that it matters much but still, that moment where you lobby to have a thread closed cause you dont agree with first post/OP. Theres a meme for that I am sure.

 

[Absolute0]

XFR is not going to boost the all core load workloads.
We have seen 6C and 8C benchmarks and they perfectly match with all core Turbo clocks for each model : 3.4Ghz MT CB run for 6C ES and ~3.5Ghz 8C MT CB run for 1700x. As for XFR it only adds 100Mhz on top of ST Turbo which means 4.1Ghz for 1800x when ST benchmark is run and 3.9Ghz for 1700x when ST benchmark is run. This is confirmed via AMD's slide and from leaked 1700X R15 ST benchmark result. AMD slide states 162 @ max clock in ST while 1700X scored 154 @ XFR bost clock of 3.9Ghz. 154*4.1/3.9=162. XFR ST boost clock for 1800X is 4.1Ghz.

All core Turbo also seems to be one bin (+100Mhz) above base clocks of each model: ~3.7Ghz for 1800X and 3.5Ghz for 1700X. That was confirmed by the numbers from official leaked AMD slide and the previous leak for 1700X. AMD slides states 1601pts for 1800X while 1700X(3.4Ghz base, 3.5Ghz ACT) scored 1537. If you adjust the clock to 3.7ghz which is supposed ACT for 1800X you will get: 1537*3.7/3.5=1624pts which is ~1.4% deviation from AMD's result and can be easily attributed to other stuff like memory clock and timings.

An excellent response, thank you, and I'm sure very informative for other readers here on new official thread as well. I didn't realize XFR is only adding 100 Mhz to the ST turbo.

I am expecting relatively limited OC headroom then. With the coolers AMD is packing, I think they are pushing these models a bit, or else they wouldn't have the "wraith max" packaged with the -X processors and just saved money packaging the "wraith spire" with everything. Also as we know when it comes to 'clocking, mo cores = mo problems, as the weakest core limits the 8-core OC. And then there's the thermals associated with 8 cores under load, which means Vcore will have to be applied very sparingly.

Conversely, since the 4-core models will be less limited by these problems, perhaps they clock past 4 Ghz easily? We will have to see how the binning affects them...

Overall, I am as impressed as everyone else, as Ryzen apparently will force Intel to cut into their 60% profit margin with a round of price cuts.

 

[PhonakV30]

This XFR does really confuse me.I don't know How works.

 

[unseenmorbidity]

I am think of getting a d15. If that cannot keep it cool, then that's insane!

 

[inf64]

An excellent response, thank you, and I'm sure very informative for other readers here on new official thread as well. I didn't realize XFR is only adding 100 Mhz to the ST turbo.

I am expecting relatively limited OC headroom then. With the coolers AMD is packing, I think they are pushing these models a bit, or else they wouldn't have the "wraith max" packaged with the -X processors and just saved money packaging the "wraith spire" with everything. Also as we know when it comes to 'clocking, mo cores = mo problems, as the weakest core limits the 8-core OC. And then there's the thermals associated with 8 cores under load, which means Vcore will have to be applied very sparingly.

Conversely, since the 4-core models will be less limited by these problems, perhaps they clock past 4 Ghz easily? We will have to see how the binning affects them...

Overall, I am as impressed as everyone else, as Ryzen apparently will force Intel to cut into their 60% profit margin with a round of price cuts.

Note that these results were likely obtained with bundled cooler AMD supplies. I suppose that ST XFR clock and maybe even ACT clock *could* be boosted more than 100Mhz IF one was to cool this chip with some great AIO WC or something even more exotic.

 

[unseenmorbidity]

Biased thread? What was wrong with the old one? Not that it matters much but still, that moment where you lobby to have a thread closed cause you dont agree with first post/OP. Theres a meme for that I am sure.

It was created as an amd bashing thread by a troll that devotes countless hours of his life to hate amd.

I bet if I go look at that old thread I will see that Fritz benchmark that was proven fake almost immediately after they were posted.

The perpetrators mostly went to troll another forum, and it is a far better environment because of it. I imagine the shame will keep them away for awhile.

EDIT: Ohh look there it is...

PS: idk why you want to stir that pot though. Just let it go.

 

[Doom2pro]

This XFR does really confuse me.I don't know How works.

It seems that it just removes the ceiling of the normal turbo boost, thermal & power willing... Basically if you turn it on with bad cooling it won't do anything above what Stock Turbo does already, however if the Tcase is low enough and the power usage is low enough it and you have it enabled, it will allow Turbo boost to go higher than it's stock limits, as far as the thermals and power let it.

 

[Ryzenenthusiast]

Ryzen 8c/16t 3.4Ghz base Validation Sample (ZD3406BAM88F4) on bench in ASRock Motherboard (X370 Gaming Fatal1ty) w/BIOS menu.

0:21 Sample Image
1:25 Heatsink going on
1:55 BIOS Menu

English Summary:

Actually I did the translation on the Youtube. So I did a longer translation on Reddit. Here it is;

Hello Everyone from Teknobiyotik.com and Teknobiyotik TV. Today we have a super product which you are awaiting for; The first working test sample of AMD Ryzen processors had arrived to our office. Now you are seeing it on my hand; Ryzen 7 1700 CPU. Along with this we have ASRock X370 Gaming K4 Motherboard in our office. We will start our system which uses this two components. We will show you the working sample of the Ryzen CPU first time in Turkey. Ryzen 7 1700 which I am holding on my hand is an equivalent for Intel i7 7700 and 7700k. Since we have some limitations forced by AMD we can not provide more details to you. But we'll show you the working sample. Let's see a working Ryzen system.

Edit: He had stated the preorders will start within couple of days on the comments section.

 

[krumme]

An excellent response, thank you, and I'm sure very informative for other readers here on new official thread as well. I didn't realize XFR is only adding 100 Mhz to the ST turbo.

I am expecting relatively limited OC headroom then. With the coolers AMD is packing, I think they are pushing these models a bit, or else they wouldn't have the "wraith max" packaged with the -X processors and just saved money packaging the "wraith spire" with everything. Also as we know when it comes to 'clocking, mo cores = mo problems, as the weakest core limits the 8-core OC. And then there's the thermals associated with 8 cores under load, which means Vcore will have to be applied very sparingly.

Conversely, since the 4-core models will be less limited by these problems, perhaps they clock past 4 Ghz easily? We will have to see how the binning affects them...

Overall, I am as impressed as everyone else, as Ryzen apparently will force Intel to cut into their 60% profit margin with a round of price cuts.

We have to remember this is 8c cpu the 6900 at 4.3 uses 210w

http://www.tomshardware.com/reviews/intel-core-i7-broadwell-e-6950x-6900k-6850k-6800k,4587-9.html

 

[Doom2pro]

the 6900 at 4.3 uses 210wl

Ouch.