First AMD Ryzen 1700X Full Review

[Shahryar_NEO]

Hi Fellas,

Here you are the full AMD Ryzen 1700X full review with complete benchmarks and comparison with rival CPUs:

http://www.shahrsakhtafzar.com/fa/re...7-1700x-review

What's your idea? Do you go with AMD again?

Enjoy

[Andi64]

Excellent CPU. Finally :')

[StAndrew]

I thought the NDA lifted tomorrow

[Ket]

"Database connection error"

[Andi64]

F5 until it loads.

If this benchmarks are true, it's an excellent CPU. Power consumption at stock speed is really low! Clock per clock is competitive with Intel, sometimes better, sometimes worse.

We now have competition! And we can finally see Bulldozer die! Finally, worst abortion of a CPU ever made (yes, even worse than Netburst) is dead!

Only problem is clock... it looks like 4.0-4.1Ghz will be max on average

[StAndrew]

I can't open due to work filters. I'm not worried about max overclock but I'm curious how AMD's 4.0 - 4.1Ghz overclock compares to Intel's 4.5+ Overclock (performance wise).

[AliG]

I got a peak at them on my phone.

Results seem to vary across the board, where Ryzen smashes some benchmarks and falls behind the pack in others. The gaming test surprised me though; didn't think Bioshock was all that CPU intensive.

[Andi64]

I can't open due to work filters. I'm not worried about max overclock but I'm curious how AMD's 4.0 - 4.1Ghz overclock compares to Intel's 4.5+ Overclock (performance wise).

I guess it will compare just like Intels 4.0Ghz vs Intels 4.5Ghz :P

Also, R7 1800X is 4.0Ghz stock, if this review is real, it looks like it's heavy binned. It took 1.45V and 100W more to get the R7 1700X to 4.0Ghz

[StAndrew]

I guess it will compare just like Intels 4.0Ghz vs Intels 4.5Ghz :P

Also, R7 1800X is 4.0Ghz stock, if this review is real, it looks like it's heavy binned. It took 1.45V and 100W more to get the R7 1700X to 4.0Ghz

Sure but I have a feeling it won't be that simple. I'm just curious, price, power and stock performance out the window, who has the absolute performance crown at max overclock. If Intel keeps the overclock performance crown, we may not see them drop prices as aggressively as we would like.

[vario]

I guess it will compare just like Intels 4.0Ghz vs Intels 4.5Ghz :P

Also, R7 1800X is 4.0Ghz stock, if this review is real, it looks like it's heavy binned. It took 1.45V and 100W more to get the R7 1700X to 4.0Ghz

Common misconception, it is NOT 4ghz stock, its 3,6ghz stock with a Turbo (not all cores) of 4ghz(4.1ghz xfr).
BTW, these results are based on ZD processor , so not retail we dont know about turbo and it was gimped with ram running at 2133 CAS 15, 7700K was running 3200.

[RPGWiZaRD]

Sure hope this isn't a good indication of what kind of OC'ing 1700X is capable of, was hoping to go AMD this round but was waiting for the final bits, overclocking, to be revealed. Hoping it would at least do ~4.5GHz, might still go Kaby Lake and shoot for a 5GHz overclock since it would save me a pretty penny too. Even the same branded motherboard in AMD flavor costs a little more than Intels as well. ASRock Z270 Taichi vs X370 Taichi, 230€ vs 270€ which also has to be added into the comparison in my case.

[xlink]

Common misconception, it is NOT 4ghz stock, its 3,6ghz stock with a Turbo (not all cores) of 4ghz(4.1ghz xfr).
BTW, these results are based on ZD processor , so not retail we dont know about turbo and it was gimped with ram running at 2133 CAS 15, 7700K was running 3200.

Does memory performance even matter? Worst case scenario you just run tighter timings, right?

[Ket]

F5 until it loads.

If this benchmarks are true, it's an excellent CPU. Power consumption at stock speed is really low! Clock per clock is competitive with Intel, sometimes better, sometimes worse.

We now have competition! And we can finally see Bulldozer die! Finally, worst abortion of a CPU ever made (yes, even worse than Netburst) is dead!

Only problem is clock... it looks like 4.0-4.1Ghz will be max on average

Thats harsh man, BD at least had alright multi-threading. Nothing will ever be worse than "nutburst".

[vario]

Does memory performance even matter? Worst case scenario you just run tighter timings, right?

Well, if youre doing slow as hell 2133Mhz with Cas-15-15-15 vs 3200 Cas 15-15-15 it does matter, albeit not much in pure computational tasks like cinebench, but in gaming it can do some damage.

@RPG wizard

You were dreaming for some time now ;-).
Remember, BDE does 4.1-4.3 on air.Also, kaby lake isnt guaranteed for 5ghz if you dont delidd.

[Andi64]

Thats harsh man, BD at least had alright multi-threading. Nothing will ever be worse than "nutburst".

This is for another debate, but IMO Netburst was an inefficient subpar architecture compared with... anything that shared it's timeline. BUT...

It had it's ups and dows. Power efficiency was really weak, on an era where nobody really cared... IPC was also weak compared even with Tualatin. But overall performance was from OK to "best in class".

It had the performance crown for some time, and when it lost it wasn't by much. You could put a P4C 3.0Ghz against an AXP 3000+ and they would trade blows.

BD, on the other hand, was the biggest failure I've seen! It was (finally, was.. :') ) an IPC nightmare, compared to Llano (or even Thuban), it was a clear step down for X6 users on many aplications. It was a disaster in power consumption, and an awfull overclocker (yeah, I know.. you could achieve 5Ghz on air... on a CPU that required 6Ghz to compete with Intel's 3Ghz). Overall performance it NEVER won on anything. It's an 8 core CPU (or 4C/8T if you want to make it sound less of a failure) that couldn't compete with Intel's i3...

I'm glad it's dead!

[AliG]

For those at work, wccftech reposted the images:
http://wccftech.com/amd-ryzen-7-1700...ck-benchmarks/

As expected, did well in most multi-thread tests - though surprisingly poorly in a few, and was fairly competitive in single threaded tests.

I'm still unsure what to make of the gaming tests. I don't have a good reason yet why some (especially Bioshock) show fairly substantial differences. Perhaps memory bandwidth?

[RPGWiZaRD]

Remember, BDE does 4.1-4.3 on air.Also, kaby lake isnt guaranteed for 5ghz if you dont delidd.

Kaby Lake has a very decent chance of reaching 5GHz what I've seen, here's a few stable overclocks on air or water copied from a swedish forum I follow:
5.0GHz 1.31v
4,95GHz 1.285v
5.0GHz 1.344v
5.0GHz 1.31v
5.0GHz 1.36v
4.8GHz 1.295v
4.8GHz 1.314v
5.0GHz 1.27v
5.0GHz 1.306v
5.0GHz 1.31v
5.0GHz 1.325v
5.0GHz 1.30v
5.0GHz 1.290v (5.1GHz 1.37v)
5.0GHz 1.34v
5.0GHz 1.30v
5.0GHz 1.36v (5.1GHz 1.385v, 5.2GHz 1.40v)
5.0GHz 1.285v
5.0GHz 1.34v

As long as you are ready to give it around 1.35~1.36v your chances should be pretty high, well 4.9GHz would still be decent too and so far I've not seen one who doesn't do 4.9GHz at least at 1.35~1.36v. 5.0GHz vs 4.1GHz for example is a 22% difference which is huge when you look at poorly threaded scenarios and if I can get the Intel significantly cheaper it makes my decision easier as I still need good one threaded performance too. But I hope this test is not 100% accurate as I would love to have more cores in the few scenarios I'll have a good use of it.

But assuming 1700X won't do much more than say 4.2~4.3GHz stable then looking at OC capability and price of the CPU + motherboard (7700k + z270 = 589€ vs 1700X + X370 = 709€), the Ryzen hype is starting to fade a bit for me and I'd see both roughly as equally good options and it'll come down to what kind of applications you are using if it's worth spending more on Ryzen over Kaby Lake (X99 platform was never interesting for me as I'm a bang for buck guy). Would explain a bit why they weren't so keen to show its overclocking capabilities just yet.

[vario]

Kaby Lake has a very decent chance of reaching 5GHz what I've seen, here's a few stable overclocks on air or water copied from a swedish forum I follow:
5.0GHz 1.31v
4,95GHz 1.285v
5.0GHz 1.344v
5.0GHz 1.31v
5.0GHz 1.36v
4.8GHz 1.295v
4.8GHz 1.314v
5.0GHz 1.27v
5.0GHz 1.306v
5.0GHz 1.31v
5.0GHz 1.325v
5.0GHz 1.30v
5.0GHz 1.290v (5.1GHz 1.37v)
5.0GHz 1.34v
5.0GHz 1.30v
5.0GHz 1.36v (5.1GHz 1.385v, 5.2GHz 1.40v)
5.0GHz 1.285v
5.0GHz 1.34v

As long as you are ready to give it around 1.35~1.36v your chances should be pretty high, well 4.9GHz would still be decent too and so far I've not seen one who doesn't do 4.9GHz at least at 1.35~1.36v. 5.0GHz vs 4.1GHz for example is a 22% difference which is huge when you look at poorly threaded scenarios and if I can get the Intel significantly cheaper it makes my decision easier as I still need good one threaded performance too. But I hope this test is not 100% accurate as I would love to have more cores in the few scenarios I'll have a good use of it.

But assuming 1700X won't do much more than say 4.2~4.3GHz stable then looking at OC capability and price of the CPU + motherboard (7700k + z270 = 589? vs 1700X + X370 = 709?), the Ryzen hype is starting to fade a bit for me and I'd see both roughly as equally good options and it'll come down to what kind of applications you are using if it's worth spending more on Ryzen over Kaby Lake (X99 platform was never interesting for me as I'm a bang for buck guy). Would explain a bit why they weren't so keen to show its overclocking capabilities just yet.

Thing is, it was never assumed that Ryzen would do 4,5-5ghz.
Other 8 cores out there arent doing it also (well maybe some 5960x on water will do 4.5 stable).
"you also agree that not all kaby lakes are doing 5.0 , moreover i am abit skeptical about what people post as "stable" , people have a tendency to be more optimistic than the reality is.
Anyhow, it was widely known that ryzen wont clock as high as kaby lake, and also about price you can show 1700 and a midrange B350 board and the price will be smaller than kaby .But yes, you probably dont reach 4.5, and most probably reach 4.1 or 4.2.
Its a simple question of what do you want, more cores or more ST speed.You cant have both.
I see it as a simple Single core or dual core problem, later on whats better E8400 or Q6600 , also how long will the platform survive.
And all this points me more towards ryzen than 7700K.

[dasa]

well the memory bandwidth is good but the latency is atrocious
2133c15
ryzen 102ns
6700k 56ns

it could explain why its performance is so far behind in a single threaded game like bioshock vs cinbench that really doesnt care so much about memory speed
hopefully this is some something amd can or has fixed

look forward to more reviews

[AliG]

well the memory bandwidth is good but the latency is atrocious
2133c15
ryzen 102ns
6700k 56ns

it could explain why its performance is so far behind in a single threaded game like bioshock vs cinbench that really doesnt care so much about memory speed
hopefully this is some something amd can or has fixed

look forward to more reviews

Yeah could be, curious to see what other's have to show as well

[Metroid]

................

I agree with you here although is too early to have a conclusion, official non-eng cpus reviews coming soon.

[dasa]

looks like all those game tests are still a little gpu limited even at low res with his gtx980
just gave bioshock a run with 6700k 2x8g@spd2133c15 gtx1070 all stock 209fps and despite gpu usage not going over 95% gtx1070 overclocked 226fps at 1366x768
edit
at 3200c14
gpu stock 220fps
gpu oc 240fps

[FlanK3r]

but you will never play at 1366x768 with highend PC....Thats the point

[chew*]

I just want to point out a key factor here.

Note memory multi is tied to northbridge frequency in his screen shots.

2133/2133.

3000/3000

Dont be surprised if you see other reviews with ram clocked i dunno say equally on both platforms that results change.

[dasa]

but you will never play at 1366x768 with highend PC....Thats the point

no nor do we need 200fps but its a good way to reduce the gpu limitation so that we dont have $50 pentiums performing the same as $500+ cpu http://www.anandtech.com/bench/CPU/1321 in gpu bottlnecked time demos which while easier to get consistent results they often dont accurately represent more demanding parts of the game
which in some cases can be gpu limited to 150fps in some places and cpu limited to 30fps in other parts of the same game
the other option is to chose games like arma (~3 threads)\total war (multithreaded) that are actually cpu limited around 60fps but the review above didnt do this although its hard to be sure without having a slower cpu in the tests like i5 7400\pentium G4560 & amd fx