Discussion AMD Ryzen MegaThread! FAQ and Resources

[eidolon171]

If the Excavator built Athlon x4 845 with a turbo clock speed of 3.8GHz scores 91.57pts in the Cinebench R15 Single-core test,

http://www.anandtech.com/show/10436/amd-carrizo-tested-generational-deep-dive-athlon-x4-845/26

and the Ryzen 7 1800X is 52% faster than Excavator, and clocked at 4.0GHz, wouldn't that produce a score of 146.5pts? That's pretty respectable. Also, what is going on with that benchmark from Anandtech where they're running everything at 3GHz?
(74.60pts/3.0GHz) x 3.8GHz ≠ 91.57pts

and what about the 4770k?
(121pts/3.0GHz) x 3.9GHz ≠ 156pts

Using the method I outline above, and assuming this,

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/18
for Ivy Bridge base

Then I calculate that the Ryzen 7 1800X CPU will be 4.4% slower than the i7 6900K in this benchmark, and 2.5% faster than the i7 3770K. That would suggest IPC Close to but not quite Haswell architecture level. Although, as was already mentioned Anandtech derived a lower score for the i7 6900K than AMD did. Why was that? Curiously, a rare Bristol Ridge CinebenchR15 score comes in at 4.3% higher than Carrizo,

http://www.anandtech.com/show/10705/amd-7th-gen-bristol-ridge-and-am4-analysis-a12-9800-b350-a320-chipset

It's probably just because of an increase in clock speed, but I don't know.

 

[salgado18]

Then I calculate that the Ryzen 7 1800X CPU will be 4.4% slower than the i7 6900K in this benchmark, and 2.5% faster than the i7 3770K. That would suggest IPC Close to but not quite Haswell architecture level.

AMD users, especially FX owners, will see an over 50% increase in IPC, so that 2-4% is irrelevant. Intel users will see no difference in IPC, honestly. They will, however, feel the difference in clocks and cores (and prices). So, being 4.4% slower than Sky Lake, in my opinion, means nothing.

But the problem is that performance is not only felt by IPC, but also by clocks. And Intel has slightly higher clocks in their lineup, which, together with higher IPC, could make stock performance better.

 

[cdrkf]

If the Excavator built Athlon x4 845 with a turbo clock speed of 3.8GHz scores 91.57pts in the Cinebench R15 Single-core test,

http://www.anandtech.com/show/10436/amd-carrizo-tested-generational-deep-dive-athlon-x4-845/26

and the Ryzen 7 1800X is 52% faster than Excavator, and clocked at 4.0GHz, wouldn't that produce a score of 146.5pts? That's pretty respectable. Also, what is going on with that benchmark from Anandtech where they're running everything at 3GHz?
(74.60pts/3.0GHz) x 3.8GHz ≠ 91.57pts

and what about the 4770k?
(121pts/3.0GHz) x 3.9GHz ≠ 156pts

Using the method I outline above, and assuming this,

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/18
for Ivy Bridge base

Then I calculate that the Ryzen 7 1800X CPU will be 4.4% slower than the i7 6900K in this benchmark, and 2.5% faster than the i7 3770K. That would suggest IPC Close to but not quite Haswell architecture level. Although, as was already mentioned Anandtech derived a lower score for the i7 6900K than AMD did. Why was that? Curiously, a rare Bristol Ridge CinebenchR15 score comes in at 4.3% higher than Carrizo,

http://www.anandtech.com/show/10705/amd-7th-gen-bristol-ridge-and-am4-analysis-a12-9800-b350-a320-chipset

It's probably just because of an increase in clock speed, but I don't know.

According to AMD's own slides in recent presentation, Ryzen 1800X scores 162 points in Cinebench R15 Single Threaded test @4.1 ghz (? Apparently that is the XFR frequency).... So it would seem in Cinebench they have exceeded the 52% IPC increase. Note that I think it's reasonable to assume the 52% would be an average- they'll likely be above and below this when looking at specific tests. That puts the IPC bang on Haswell (so needs a small clock speed advantage to match broadwell- which is how the 1800X matches 6900k).

Also just saw this, it's is from wtfbbqtech so take as you will but would be amazing if true...
http://wccftech.com/amd-ryzen-7-1800x-world-record/

 

[$hawn]

So I'm really confused about AMD's XFR, especially after watching Linus' video.

Does XFR only boost the clock above max turbo on a single core by a meagre 100MHz? Seems too little for a feature they advertised so much about. Or do they mean you'd only get that much with the stock cooler?

I previously thought that the clock would boost as high as is possible depending on how good your cooling was - something like an automatic clocking up to 4.6GHz on a single core would have been be possible on water cooling, if power was still under limits.

 

[ScrewySqrl]

as of right now ryzen CPUS are #1, #3 and #4 best selling CPUs on Amazon:

(image taken at 8:33 eastern time this morning)

 

[juanrga]

I completely agree also unity can really make use of all the cores i'd always recommend a 8 core fx over a I3 skylake in that case, or with virtual machines just not games. Ryzen man i'm so happy and i can't tell people enough in this thread how much i'm happy to be wrong and in a way Juan i feel thinks the same.

Underestimating and getting more then what i expected is a far better feeling then i had with bulldozer where i was expecting big and got disappointed.

Of course! I prefer to be accurate, but I am going to fail I prefer to underestimate rather than overestimate. I am glad that Ryzen is about 5--10% better than I expected. But look at the other extreme, look at all that people that predicted 12-cores, quad-channel AM4, 4.0GHz base clocks and 4.5GHz turbo, SKL/KBL IPC, 5GHz OC on air, 256bit FMAC units,... They overestimated and hyped Ryzen from 11% up to 100%.

 

[juanrga]

If the Excavator built Athlon x4 845 with a turbo clock speed of 3.8GHz scores 91.57pts in the Cinebench R15 Single-core test,

http://www.anandtech.com/show/10436/amd-carrizo-tested-generational-deep-dive-athlon-x4-845/26

and the Ryzen 7 1800X is 52% faster than Excavator, and clocked at 4.0GHz, wouldn't that produce a score of 146.5pts? That's pretty respectable. Also, what is going on with that benchmark from Anandtech where they're running everything at 3GHz?
(74.60pts/3.0GHz) x 3.8GHz ≠ 91.57pts

and what about the 4770k?
(121pts/3.0GHz) x 3.9GHz ≠ 156pts

Using the method I outline above, and assuming this,

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/18
for Ivy Bridge base

Then I calculate that the Ryzen 7 1800X CPU will be 4.4% slower than the i7 6900K in this benchmark, and 2.5% faster than the i7 3770K. That would suggest IPC Close to but not quite Haswell architecture level. Although, as was already mentioned Anandtech derived a lower score for the i7 6900K than AMD did. Why was that? Curiously, a rare Bristol Ridge CinebenchR15 score comes in at 4.3% higher than Carrizo,

http://www.anandtech.com/show/10705/amd-7th-gen-bristol-ridge-and-am4-analysis-a12-9800-b350-a320-chipset

It's probably just because of an increase in clock speed, but I don't know.

According to AMD's own slides in recent presentation, Ryzen 1800X scores 162 points in Cinebench R15 Single Threaded test @4.1 ghz (? Apparently that is the XFR frequency).... So it would seem in Cinebench they have exceeded the 52% IPC increase. Note that I think it's reasonable to assume the 52% would be an average- they'll likely be above and below this when looking at specific tests.

Exactly! According to AMD internal data the IPC gap over Excavator is 58% on Cinebench R15.

 

[Abo_3_]

Guys.. Chill...

https://www.youtube.com/watch?v=rw7SkWnSFug

 

[truegenius]

Well in terms of overclocking.. yes, I spotted this on a HardForum thread (where he/she spotted it from anandtech).

https://hardforum.com/threads/amd-announces-the-official-ryzen-launch-date.1925523/page-4#post-1042833782

Looks like we might be getting some ability to overclock on each individual core.

overclocking individual cores was available on amd overdrive from as early as year 2012 ( overdrive in my driver cd ) , i use different clocks on different cores as core 4,5 needs too much voltage than other cores, cores 1,2,3 are golden cores, core 6 is in between but raises cpu temp more than all other 5 cores.

 

[truegenius]

jdwii said:
I completely agree also unity can really make use of all the cores i'd always recommend a 8 core fx over a I3 skylake in that case, or with virtual machines just not games. Ryzen man i'm so happy and i can't tell people enough in this thread how much i'm happy to be wrong and in a way Juan i feel thinks the same.

Underestimating and getting more then what i expected is a far better feeling then i had with bulldozer where i was expecting big and got disappointed.

looking at the cb15 score zen is 1% slower than haswell ipc, 54% faster than phenom 2 ipc, 72% faster than piledriver ipc, 8% faster than ivy bridge ipc
mine was ivy level ipc ±5% in cb11.5
if i consider same trend in cb11.5 then zen performed ~3% better than my expectation (when considering error margin too )

btw we got best of both world, haswell ipc and 45% smt boost ( almost exactly my prediction of 1.75-2x better than intel and near 50% )

now left to see is overclocking headroom, any guestimation ?
to me it looks like 4-4.2ghz will be the limit on air on all cores.
But if they can make 4.5ghz on 1-4 cores then 4ghz all core won't be that bad then.

btw on topic of ram speed, this article says that new games are taking advantage of faster ram, so lowest latency and fastest speed at reasionable price will be good option ( also will make for low overclocking headroom of zen )
http://www.eurogamer.net/articles/digitalfoundry-2016-is-it-finally-time-to-upgrade-your-core-i5-2500k

 

[8350rocks]

http://cpu.userbenchmark.com/Compare/AMD-Ryzen-7-1700X-Turbo-Disabled-vs-Intel-Core-i7-4770K/m243187vs1537

4770K vs 1700X with turbo disabled. 4770K is ahead by <1% in single core.

 

[eidolon171]

According to AMD's own slides in recent presentation, Ryzen 1800X scores 162 points in Cinebench R15 Single Threaded test @4.1 ghz (? Apparently that is the XFR frequency).... So it would seem in Cinebench they have exceeded the 52% IPC increase. Note that I think it's reasonable to assume the 52% would be an average- they'll likely be above and below this when looking at specific tests. That puts the IPC bang on Haswell (so needs a small clock speed advantage to match broadwell- which is how the 1800X matches 6900k).

Also just saw this, it's is from wtfbbqtech so take as you will but would be amazing if true...
http://wccftech.com/amd-ryzen-7-1800x-world-record/

Well reasoned.

 

[wh3resmycar]

so march for the R7s. anybody know when the R5s gonna hit?

Nope. I was kinda bummed that all they showed was the R7s. I still wonder when R5s are gona come.

yeah it was a shame. i'm pretty sure a R5 1600x OC'd to 4.5/4.6ghz will make me forget about any lga 1151 i7 much less an i5. it could probably even beat it's bigger r7 brethren assuming it will OC higher due to less cores/heat in some workloads.

 

[jaymc]

According to AMD's own slides in recent presentation, Ryzen 1800X scores 162 points in Cinebench R15 Single Threaded test @4.1 ghz (? Apparently that is the XFR frequency).... So it would seem in Cinebench they have exceeded the 52% IPC increase. Note that I think it's reasonable to assume the 52% would be an average- they'll likely be above and below this when looking at specific tests. That puts the IPC bang on Haswell (so needs a small clock speed advantage to match broadwell- which is how the 1800X matches 6900k).

Also just saw this, it's is from wtfbbqtech so take as you will but would be amazing if true...
http://wccftech.com/amd-ryzen-7-1800x-world-record/

Well reasoned.

wowzers...

 

[MA2K]

how well do you think a ryzen 5 will be for gaming compared to 6/7th gen low-mid tier i7s?

 

[thegentlewoman]

Exactly, that is a supposition (on 1600x) also some people have made me wonder about. That is why it's a great idea to wait for the non filled pocket or ppl that are building the machine piece by piece like me. (juanrga was suggesting it as many other wise ppl, not me)

I wanna say thanks for the asnwers on the other page, I really appreciate and learned more. Now we get why there is a "sweet spot" in the price of RAM, it's just that they are 16 CL or 19 CL so to a 14 CL that has inferior frequency that matters (and can be "read" on the price first of all).

I wanna share an info, since I saw somebody writing about a UK retailer prices: that shop from Danemark also posted today the 3 top of the line Zen CPU: It's a 379 499 and 579 Euro... and ... it's a lot.
An other (or a first reason, at least to me) to wait Reviews, and wait the 1600X .. since I wanted to get the 1700, but, hell, no (maybe later in 2 years reselling this first CPU I will buy, or maybe there will be no need because of the XFR possibilities).
Many think that dissipation and logic will work the same, colder and stable means more OC possible. But, is there a chance that these CPU just work differently? Maybe simply not allowing such a bug OC? and so just sticking to "a limited" one that Linus for example mentions?

I am happy for the prices and various choices of X370: but the top of gamma are really too "Exuberant" (too much stuff I don't need) so maybe I will go for a Prime Asus or if the 1600X is not too pricey and if the reviews will "sing" accordingly a VI Hero MOBO (which is clearly very tasty).

The advice and the "read" of this job and tech I saw on one of the early reviews suggestion video is to choose a good CPU and then get the other HW accordingly, not base (as we have done for the last 20 years) everything starting from a good MOBO (and then coordinate all the rest around it).

Anyway it's a huge revolution, the video with Streetfighter music ,for us born before the 90ties, is really funny...
Paparazzi flash are EXACTLY THOSE you can see in the SNES game... I still play it with my phone and a nintendo bloothooth chinese spin-off joypad.

Viva Taipei and, what to say... I am more excited for April and especially for this summer with incredible Vega and then so called "Navi" GPUs coming out - 2018- (but already the 2017 summer 16 gb and +++ transistor stuff coming out of AMD ..and I mean, that means a lot: lower TDP requested, no anymore stressing ourselves with not working SLi\crossfires stuff and also possibility to see definitely if VR will become a commodized technology or not (these GPU will definitely mass enable its use ...

In terms of singularity means a lot... (some people will even go physically mine their Bitcoins with VR Videogames .. just for fun) ... LOL.
Sorry for long post.

Wanna see the prices in this EU website (It's just randomly again no ads ment here, I don't know them.. it's just curious\info): Check here then https://www.proshop.de/?s=Ryzen

 

[Nope 1151]

search.jd.com/search?keyword=ryzen&enc=utf-8&qrst=unchange&rt=1&vt=2&ev=exbrand_AMD%40&uc=0#J_searchWrap\
422 USD for 1700x, 571 USD for 1800x, 357 for 1700
Intel 6900k is 1171 USD here
Ryzen is 51% below intel. No tax here.
Says Packaging with cooler is 0.5Kg

 

[Hellfire13]

Seems pretty interesting so far. Question is, how will they fare on gaming benchmarks.
On another note, there is a rumour of a i7-7740K and a i5-7640K to counter Ryzen... https://videocardz.com/65708/intel-preparing-core-i7-7740k-a-response-to-amd-ryzen-7-1800x

 

[nikoli707]

If the Excavator built Athlon x4 845 with a turbo clock speed of 3.8GHz scores 91.57pts in the Cinebench R15 Single-core test,

http://www.anandtech.com/show/10436/amd-carrizo-tested-generational-deep-dive-athlon-x4-845/26

and the Ryzen 7 1800X is 52% faster than Excavator, and clocked at 4.0GHz, wouldn't that produce a score of 146.5pts? That's pretty respectable. Also, what is going on with that benchmark from Anandtech where they're running everything at 3GHz?
(74.60pts/3.0GHz) x 3.8GHz ≠ 91.57pts

and what about the 4770k?
(121pts/3.0GHz) x 3.9GHz ≠ 156pts

Using the method I outline above, and assuming this,

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/18
for Ivy Bridge base

Then I calculate that the Ryzen 7 1800X CPU will be 4.4% slower than the i7 6900K in this benchmark, and 2.5% faster than the i7 3770K. That would suggest IPC Close to but not quite Haswell architecture level. Although, as was already mentioned Anandtech derived a lower score for the i7 6900K than AMD did. Why was that? Curiously, a rare Bristol Ridge CinebenchR15 score comes in at 4.3% higher than Carrizo,

http://www.anandtech.com/show/10705/amd-7th-gen-bristol-ridge-and-am4-analysis-a12-9800-b350-a320-chipset

It's probably just because of an increase in clock speed, but I don't know.

im lookin for it but i cant find it, but i saw 4.1ghz@149 cinebench r15 single core. if i find it i will put it up.

im also very interested like others are in the 1600x and its overclock scaling compared to its 8 core siblings. if it can overclock better then it may be an excellent chip. reading puget reviews they many times come across programs that simply have very poor performance scaling past 8-12-16 threads. i will have to see reviews but i would trade 16 threads for 6c12t if i get 80-85% of the 8c16t performance while giving me 110% of the single core performance. there is a trade off there that for the time being will benefit gaming and overall average desktop snappiness.

 

[jwcrellin]

Okay, right now intel has like a 70% desktop pc cpu market share. What will that be down to by this time next year????

 

[Nope 1151]

We don't know. There is so much uncertainty about how Intel will respond it seems (Even in their own ranks!). For all we know, they may resort to underhand practices again.

Von Clausewitz always said that war is inherently chaotic and unpredictable, and only fools would apply the same theory to every situation. It is not different in businesses.

But if Ryzen is good? Anyone who got burned by Bulldozer and switched + The current Vishera users + Anyone with Sandy/Ivy CPU or earlier is the probable at best guess. And maybe some people salty about Intel's current Gen.

 

[jwcrellin]

http://www.guru3d.com/news-story/amd-ryzen-5-1600x-cinebench-r15-score-close-to-intel-core-i7-6800k.html
This will be a killer vs the 6800k

 

[jdwii]

Being completely honest i don't see much coming except maybe their X99 replacement but do people think they are really going to just cut everything down with the current products. X99 boards are expensive and those would have to come down in price as well. I saw decent AM4 boards going for 100$ that can put a 1800X in it.

 

[jwcrellin]

Yes, x299 will debut in August, it will be interesting to see what features and extras intel adds to that platform now that ryzen has shaken things up.

 

[nikoli707]

all of this isn't news to intel, im sure intel has already torn apart multiple ryzen cores to see whats inside long before we even knew the name change from zen to ryzen. intel will answer but we dont know what that answer will be. price drops? new product? who knows. for now we should all rejoice because no matter what happens we are getting more performance at a lower price.