Discussion AMD Ryzen MegaThread! FAQ and Resources

[juanrga]

You mean the chip that has a CPU Mark of 1825 @ 3.4 GHz + turbo? In other words, 100 points lower clock per clock?

No. I mean the SB i5 that does 2428 @4.5Ghz and thus about 2050 @3.8GHz, whereas RyZen does 2046 @3.8GHz.

 

[8350rocks]

You mean the chip that has a CPU Mark of 1825 @ 3.4 GHz + turbo? In other words, 100 points lower clock per clock?

No. I mean the SB i5 that does 2428 @4.5Ghz and thus about 2050 @3.8GHz, whereas RyZen does 2046 @3.8GHz.

I will say wait for the benchmarks. For all we can see, Ryzen is running at 3.4 GHz...which puts it 200 points ahead of SB clock per clock.

EDIT: Another set of CPU Mark benchmarks show an overclocked APU with turbo disabled and turbo says N/A. I am pretty confident that is without turbo, to be honest.

EDIT2: PROOF! The MB used in the 1700X CPU Mark tests was an A320 MB, an entry board that does not support XFR turbo. This is why turbo is disabled, and why the board did not pick up a turbo reading. Because turbo was disabled by the hardware since XFR does not work on the A320 chipset.

Ryzen CPU Mark scores are at 3.4 GHz.

 

[juanrga]

You mean the chip that has a CPU Mark of 1825 @ 3.4 GHz + turbo? In other words, 100 points lower clock per clock?

No. I mean the SB i5 that does 2428 @4.5Ghz and thus about 2050 @3.8GHz, whereas RyZen does 2046 @3.8GHz.

I will say wait for the benchmarks. For all we can see, Ryzen is running at 3.4 GHz...which puts it 200 points ahead of SB clock per clock.

EDIT: Another set of CPU Mark benchmarks show an overclocked APU with turbo disabled and turbo says N/A. I am pretty confident that is without turbo, to be honest.

I demonstrated turbo is enabled here

http://www.tomshardware.co.uk/forum/id-3327589/amd-ryzen-megathread-faq-resources/page-2.html#19284340

Moreover, this test agrees with CPCHardware results for multithreaded performance

i7-6850k < RyZen 1700X < i7-6900k

 

[8350rocks]

You mean the chip that has a CPU Mark of 1825 @ 3.4 GHz + turbo? In other words, 100 points lower clock per clock?

No. I mean the SB i5 that does 2428 @4.5Ghz and thus about 2050 @3.8GHz, whereas RyZen does 2046 @3.8GHz.

I will say wait for the benchmarks. For all we can see, Ryzen is running at 3.4 GHz...which puts it 200 points ahead of SB clock per clock.

EDIT: Another set of CPU Mark benchmarks show an overclocked APU with turbo disabled and turbo says N/A. I am pretty confident that is without turbo, to be honest.

I demonstrated turbo is enabled here

http://www.tomshardware.co.uk/forum/id-3327589/amd-ryzen-megathread-faq-resources/page-2.html#19284340

Moreover, this test agrees with CPCHardware results for multithreaded performance

i7-6850k < RyZen 1700X < i7-6900k

Look at my second edit, the proof is in the hardware used to conduct the test. The MB does not support XFR turbo, so turbo is disabled.

 

[ex_bubblehead]

And we've just had our first involuntary departure from the building. A reminder that this thread will be moderated even tighter than the old thread. There is a zero tolerance policy in effect for uncivil conduct and direct calling out of other members.

 

[Nope 1151]

Has anyone done a comparison to the CPC leaks? I am not (mathematically) competent enough for that.

This Passmark bench puts 8C Ryzen 3.4/3.8GHz average multithreading performance somewhat between 6C and 8C Broadwell, and the former CPC leak did the same

Thanks mate! Hopefully there is no hype train here.

 

[jdwii]

Couple things that have me a bit concerned about those results from videocardz, one is we don't know with absolute evidence that turbo is on or off. I doubt that turbo would be off just cause its a A320M as i don't recall Amd claiming even the 300 series boards would have to have turbo off. BUT i'm not convinced its on either cause we might not and probably don't have the latest BIOS.

Slightly less concerning is that it is using the lower mainstream board from Amd and if you look at the ram its not even using the same timings or settings.

All these things can make a decent difference all we know for sure is that Ryzen IMO is not a failure and yeah at times its beating a 6900K and that is with accounting for the information above, if we can get this CPU with 6800K prices i can say Amd has a nice product on their hands.

If these results are 100% final and i don't think they are i can't see Amd asking more then 500$ to keep their advantage on price/performance.

 

[8350rocks]

Couple things that have me a bit concerned about those results from videocardz, one is we don't know with absolute evidence that turbo is on or off. I doubt that turbo would be off just cause its a A320M as i don't recall Amd claiming even the 300 series boards would have to have turbo off. BUT i'm not convinced its on either cause we might not and probably don't have the latest BIOS.

Slightly less concerning is that it is using the lower mainstream board from Amd and if you look at the ram its not even using the same timings or settings.

All these things can make a decent difference all we know for sure is that Ryzen IMO is not a failure and yeah at times its beating a 6900K and that is with accounting for the information above, if we can get this CPU with 6800K prices i can say Amd has a nice product on their hands.

If these results are 100% final and i don't think they are i can't see Amd asking more then 500$ to keep their advantage on price/performance.

A few retailers are already stocking parts, the flagship is listed for $499.99 in one place, and $491 in another.

EDIT: The part we see benchmarked here is not the flagship, that part is listed for ~$400, IIRC.

 

[Cazalan]

Look at my second edit, the proof is in the hardware used to conduct the test. The MB does not support XFR turbo, so turbo is disabled.

If it does that well on an A series MB w/bad RAM timings and no turbo, can't wait to see it on an X series w/water. Exciting times.

 

[Nope 1151]

Side question, is there a necessity to upgrade these days? I don't game, but I do code and browse the web, how different would Ryzen compare to the Phentom II X6 1100T?

On a more serious note, many of you bring up the entry level MB issue, do you mean that is probably "the lowest" that chip could go?

 

[jdwii]

Side question, is there a necessity to upgrade these days? I don't game, but I do code and browse the web, how different would Ryzen compare to the Phentom II X6 1100T?

On a more serious note, many of you bring up the entry level MB issue, do you mean that is probably "the lowest" that chip could go?

Basically meaning that the board is not as high-end as a X99 platform just not 100% fair. Also if you just browse the web and code(hey you work with AI my friend is basically maxing out his 8GB of ram doing that ha ha) all i can personally talk for is basic PC usage and browsing.

From a performance perspective i see no reason to upgrade, however for features such as USB 3.1, low temps which means low noise, low wattage, newer M,2 support,DDR4 memory, if none of that matters to you then i see little reason to upgrade i owned a Phenom II 1100T was a amazing processor loved it. Personally i'd still upgrade to a modern platform but its by far not needed. EDIT a SSD would be a massive upgrade if you only have a HD.

 

[8350rocks]

Side question, is there a necessity to upgrade these days? I don't game, but I do code and browse the web, how different would Ryzen compare to the Phentom II X6 1100T?

On a more serious note, many of you bring up the entry level MB issue, do you mean that is probably "the lowest" that chip could go?

Ryzen would speed up compile times by a good margin, not sure how often you compile, but that could make a significant difference.

Other than that...unless you render or something...you are probably good.

 

[Nope 1151]

Side question, is there a necessity to upgrade these days? I don't game, but I do code and browse the web, how different would Ryzen compare to the Phentom II X6 1100T?

On a more serious note, many of you bring up the entry level MB issue, do you mean that is probably "the lowest" that chip could go?

Ryzen would speed up compile times by a good margin, not sure how often you compile, but that could make a significant difference.

Other than that...unless you render or something...you are probably good.

I see... Are hand me down PCs' a thing? Cause mine is.

 

[nikoli707]

is there any 4c8t benchmarks? can we assume the 4c8t chip is the same as the 8c16t chip? if it is the same then a 4c8t chip will run much cooler and provide a much higher overclock ceiling.

lets figure a cpu+cooler+mobo+ram price is always paired... a 7700k build will cost $570 (optimistic prices of $320+$50+$120+$80)

if the ryzen 4c8t chip can overclock nicely and can hit 85% of the overall performance of a 7700k@4.8ghz(both using the same cpu cooler/case/build/etc) and do it at a $210+$50+$110+$80=$450 then i think intel will have a problem on its hands. if it costs lower then great. all amd needs to do is hit those benchmarks all around.

the same story goes for the 6800k and the 6c12t ryzen variant except this is where im much more optimistic for amd. 6800k package of $390+$50+$180+$80=$700.... if a ryzen package costs $550 and hits 85% performance of the 6800k@4.2ghz then its a grand slam, effectively making the 7700k package a very hard sell at current prices.

we will have to see how it all plays out. pricing is king here and without knowing chipset and cpu pricing its all just a mystery.

 

[illegaloperation]

Have the processors and motherboards been finalized and handed out to reviewers?

Obviously, they would NDA.

 

[juanrga]

You mean the chip that has a CPU Mark of 1825 @ 3.4 GHz + turbo? In other words, 100 points lower clock per clock?

No. I mean the SB i5 that does 2428 @4.5Ghz and thus about 2050 @3.8GHz, whereas RyZen does 2046 @3.8GHz.

I will say wait for the benchmarks. For all we can see, Ryzen is running at 3.4 GHz...which puts it 200 points ahead of SB clock per clock.

EDIT: Another set of CPU Mark benchmarks show an overclocked APU with turbo disabled and turbo says N/A. I am pretty confident that is without turbo, to be honest.

I demonstrated turbo is enabled here

http://www.tomshardware.co.uk/forum/id-3327589/amd-ryzen-megathread-faq-resources/page-2.html#19284340

Moreover, this test agrees with CPCHardware results for multithreaded performance

i7-6850k < RyZen 1700X < i7-6900k

Look at my second edit, the proof is in the hardware used to conduct the test. The MB does not support XFR turbo, so turbo is disabled.

I am reading the second edit now. That is not true.

First, I demonstrated that the performance gap is the same (+- 1%) in multithreading that in single-thread. If turbo was not working the gap would increase a lot of in the single-thread test because all other chips would be pushing to single-core turbo clocks. But this is not happening. The conclusion is that turbo is enabled on RyZen.

Second, all Ryzen boards support stock turbo. Ryzen defines two single-core turbos: F_TST and F_TMAX. The first is the stock turbo. It is 3.8GHz for the 1700X model. This turbo is the analogous of the 4.2GHz of the FX-8350 or the analogous of the 3.9GHz of the i5-6600k. As stated above the stock turbo is enabled on any board.

Ryzen has a second turbo mode, the F_TMAX mentioned above. This is an overclocking turbo. This is the famous XFR. This is an variable turbo, not defined at fabric, because it depends on parameters measured by the chip such as temperature. It is this overclocking turbo which is not supported by all the mobos. You need a mobo with overclocking support. i.e. a mobo with XFR support.

The mobo used in the Passmark testing doesn't support XFR; therefore, the F_TMAX turbo was disabled, but the chip still run at stock turbo, as demonstrated above. In short:

1700X base clock: 3.4GHz
1700X single-core default turbo: 3.8GHz
1700X single-core overclocking turbo: ? <-- it depends on cooling installed, the quality of the XFR-enabled mobo, and the silicon lottery.

 

[juanrga]

Full Ryzen family

 

[salgado18]

Full Ryzen family

Two conclusions, then:

1. terrible naming again, AMD? Why not use the 9 family for 8c chips, 7 for 6c, 5 for 4c/8t, and 3 for 4c/4t? And what's the deal with the PRO and X? Where are the guys who named the RX 4xx family? What's going on?

2. Ryzen will attack almost every Intel with one socket, from low-end (i3) to very high-end (8c i7). I am very, very pleased with that.

Also question: If it's possible that there is a 6c chip, can there be a 2c chip for the very-low-end? Or will they push the cheapest cpus into quad-cores, killing the dual-cores from the market?

 

[juanrga]

Full Ryzen family

Two conclusions, then:

1. terrible naming again, AMD? Why not use the 9 family for 8c chips, 7 for 6c, 5 for 4c/8t, and 3 for 4c/4t? And what's the deal with the PRO and X? Where are the guys who named the RX 4xx family? What's going on?

2. Ryzen will attack almost every Intel with one socket, from low-end (i3) to very high-end (8c i7). I am very, very pleased with that.

Also question: If it's possible that there is a 6c chip, can there be a 2c chip for the very-low-end? Or will they push the cheapest cpus into quad-cores, killing the dual-cores from the market?

"Pro" are professional products. AMD has currently several Pro chips like the A10 PRO-7800B, a Kaveri-based APU for bunsiness.

I am not sure about distinction between X and non-X. I guess X models support XFR.

If there is any dual core CPU, it would be a Raven Ridge APU with half cores disabled and iGPU disabled.

 

[8350rocks]

Full Ryzen family

Two conclusions, then:

1. terrible naming again, AMD? Why not use the 9 family for 8c chips, 7 for 6c, 5 for 4c/8t, and 3 for 4c/4t? And what's the deal with the PRO and X? Where are the guys who named the RX 4xx family? What's going on?

2. Ryzen will attack almost every Intel with one socket, from low-end (i3) to very high-end (8c i7). I am very, very pleased with that.

Also question: If it's possible that there is a 6c chip, can there be a 2c chip for the very-low-end? Or will they push the cheapest cpus into quad-cores, killing the dual-cores from the market?

The yields would have to be abysmal for there to be a dual core enthusiast CPU, and they are not abysmal.

I doubt we see dual cores outside of potential mobile SKUs.

 

[ScrewySqrl]

Full Ryzen family

Two conclusions, then:

1. terrible naming again, AMD? Why not use the 9 family for 8c chips, 7 for 6c, 5 for 4c/8t, and 3 for 4c/4t? And what's the deal with the PRO and X? Where are the guys who named the RX 4xx family? What's going on?

the numbers roughly match what they compete against.

3=i3 (x3xx)
5=i5 (x5xx/x6xx)
7=i7 quad cores (x7xx)
9=i7 6/8 cores (x9xx)

 

[illegaloperation]

1. terrible naming again, AMD? Why not use the 9 family for 8c chips, 7 for 6c, 5 for 4c/8t, and 3 for 4c/4t? And what's the deal with the PRO and X? Where are the guys who named the RX 4xx family? What's going on?

AMD is adopting Intel's naming scheme.

Ryzen 3 ~ Core i3
Ryzen 5 ~ Core i5
Ryzen 7 ~ Core i7

For example:

Ryzen 7 1800X

Ryzen 7 = high end

1 = first generation

8 = high end processor (highest 9, lowest 0)

00 = used to differentiate similar products that might have slightly different clocks (higher number = higher clock)

X = extended frequency range

 

[salgado18]

Especially because they copied Intel, the numbers are still terrible. Look:

- Is the 1800 better than the 1700? Yes, but kind of, because it's probably just clocks, and usually you could go with the 1700 and overclock, but it's a different series;

- Or just take a Ryzen 5, which is a great mid-range right? Sort of, because the 5 1600 and 1500 are 6-cores, and the 1400 and 1300 are 4-cores.

So the 3, 5 and 7 mean different things, and you need to look up, and eventually memorise, the model number, just like Intel's i7 series (which is worse, because it changes sockets with the same CPU name).

Compare that to FX 8-series (8 cores), 6-series and so on, also 8120 = first gen, 8320 = second gen, and done, you know exactly what you are getting from the numbers.

I don't say it makes no sense, I just mean it could be way, way better.

 

[010TheMaster010]

http://www.tweaktown.com/news/56234/amd-ryzen-leaks-r7-1800x-costs-499-line-up-strong/index.html
If this sticks then that 1400X is looking like my new best friend

 

[jdwii]

http://www.tweaktown.com/news/56234/amd-ryzen-leaks-r7-1800x-costs-499-line-up-strong/index.html
If this sticks then that 1400X is looking like my new best friend

Now those prices make a lot more sense then what i was seeing before, people where spreading crap like quad cores competing with Pentiums those rumors i even hoped were false.