Ryzen "2" ?

humbug · 26 Mar 2018 at 13:16

Zeed said:
And Benchmarks are out
https://i.imgur.com/2vm5fZk.jpg?1
https://imgur.com/Y28Zmr8

HeX said:
They ran it on a super budget A320, so no overclocking, and no XFR 2.0. Expect launch reviews on 470/450 boards to show additional performance.

Then this is almost purely IPC given that the 1800X runs at 3.6Ghz and the 2700X at 3.7Ghz.

Given the clock speed differences between the 8700K and the 2700X (3.7 vs 4.3) just 14% difference is hugely impressive, when XFR2 is working it will cut that difference down even more.

HeX · 26 Mar 2018 at 13:17

sideways14a said:
Just saw that, so no xfr... that will be interesting once it starts working - hoping for more than a 50mhz improvement this time what with the higher power available.

Old XFR boost was dependent on the cpu some had a 50MHz XFR boost, others like the 1500X had a 200MHz boost.

The big thing for the new XFR is it allows boosting of >2 cores, and scales gradually with core usage and thermal/power headroom.

sideways14a · 26 Mar 2018 at 13:19

Might see 4.5 Big ones yet .... maybe.

NinjaCool · 26 Mar 2018 at 13:22

Using a budget A320 board on a Ryzen 2 CPU is not exactly ideal.
I was hoping this review would use one of the new 470 boards with fast DDR4 ram, so we could see the max potential of the Ryzen 2.

humbug · 26 Mar 2018 at 13:31

humbug said:
Then this is almost purely IPC given that the 1800X runs at 3.6Ghz and the 2700X at 3.7Ghz.

Given the clock speed differences between the 8700K and the 2700X (3.7 vs 4.3) just 14% difference is hugely impressive, when XFR2 is working it will cut that difference down even more.

Forgot to add, the 1800X has a base clock of 3.6Ghz, all core boost of 3.7Ghz, a single core boost of 4Ghz and XFR of 4.1Ghz.

Depending on what that A320 board is doing with that 2700X its upto 4% IPC gain in gaming, i suspect that board is running a test BIOS or even Ryzen 1### BIOS and isn't boosting at all, so i can see a large chunk of that 14% gaming advantage the 7700/8700K has evaporate once the official benchmarks are out.

DragonQ · 26 Mar 2018 at 13:45

The 12 nm node certainly provides no power usage improvements. As expected it's an overclocked refresh, a la Radeon Rx 5xx series. I wouldn't expect IPC changes except for supporting faster RAM but speculating on IPC based on those graphs is foolish since no clock references are given.

NinjaCool · 26 Mar 2018 at 13:55

Videocardz has some info up on the site.
https://videocardz.com/75504/cpc-hardware-amd-ryzen-7-2700x-review-goes-online

humbug · 26 Mar 2018 at 13:56

DragonQ said:
The 12 nm node certainly provides no power usage improvements. As expected it's an overclocked refresh, a la Radeon Rx 5xx series. I wouldn't expect IPC changes except for supporting faster RAM.

Ryzen is very efficient anyway, more efficient than Intel so i really don't care even if it uses a bit more power.

Don't be so sure about IPC, the 2600X is running near identical base clock as the 2700X, the gaming performance is near identical, you're looking at these figures without them boosting anything, both are faster than the 1800X which it boosting, possibly even running higher clocks.

I'm to a limited extent encouraged by this, especially looking at the difference to Intel's finest mainstream, 14% is not a lot and this slide looks very (4 core or less) i would be happy if its stays like that but i do think once Ryzen 2### is on an X or B350/70 board Intel's gaming lead will be reduced to single figures.

DragonQ · 26 Mar 2018 at 14:04

humbug said:
Ryzen is very efficient anyway, more efficient than Intel so i really don't care even if it uses a bit more power.

Ryzen is rather efficient but only up to a point (mid-3 GHz region IIRC). It's clear that they are going balls to the wall here by bumping the clocks without maintaining a power envelope. I don't think that's necessarily a bad decision, just a fact.

humbug said:
Don't be so sure about IPC, the 2600X is running near identical base clock as the 2700X, the gaming performance is near identical, you're looking at these figures without them boosting anything, both are faster than the 1800X which it boosting, possibly even running higher clocks.

Sorry, how do you know the 1800X is boosting and the 2xxx chips are not? Is it mentioned in the French text? The A320 does apply turbo boost to CPUs as far as I know. It sounds like you're just assuming this is the case.

humbug · 26 Mar 2018 at 14:12

DragonQ said:
Ryzen is rather efficient but only up to a point (mid-3 GHz region IIRC). It's clear that they are going balls to the wall here by bumping the clocks without maintaining a power envelope. I don't think that's necessarily a bad decision, just a fact.

Sorry, how do you know the 1800X is boosting and the 2xxx chips are not? Is it mentioned in the French text? The A320 does apply turbo boost to CPUs as far as I know. It sounds like you're just assuming this is the case.

If the 1800X isn't boosting then we would have to assume its broken, why would it not be boosting?

I don't read French but a couple of people in this thread have said its not running XFR. if the 2700X is the IPC is less than Ryzen 1### because its XFR is 7% higher than the 1800X. the IPC is obviously not going to be lower, well it might be but if it is then AMD had better not release it at all.

DragonQ · 26 Mar 2018 at 14:17

Ah here we go, more detail in the Videocardz link. R7 1800X would be running at 3.7 GHz all cores, whereas R7 2700X would be 3.9 GHz, so a 5.5% clock speed bump. The non-gaming benchmarks show a ~9% improvement; I imagine the difference is due to improved memory latency although again this is just speculation. Of course for the gaming benchmarks we have no idea how many cores are in use so it could be literally anywhere between 3.9 and 4.3 GHz for each individual game.

TrixP10 · 26 Mar 2018 at 14:21

humbug said:
If the 1800X isn't boosting then we would have to assume its broken, why would it not be boosting?

I don't read French but a couple of people in this thread have said its not running XFR. if the 2700X is the IPC is less than Ryzen 1### because its XFR is 7% higher than the 1800X. the IPC is obviously not going to be lower, well it might be but if it is then AMD had better not release it at all.

Well I think this a scammy pre review to get headlines and preempt any good press. The French are not well known for buying AMD and I don't think they worry too much about NDA's. What they have released is a preview of an 2700x shown in as bad a light as possible.
When has that happened before?
Soon there will be some better info - hopefully.

humbug · 26 Mar 2018 at 14:24

Hopefully its more fake crap then because ^^^^ ....

DragonQ said:
Ah here we go, more detail in the Videocardz link. R7 1800X would be running at 3.7 GHz all cores, whereas R7 2700X would be 3.9 GHz, so a 5.5% clock speed bump. The non-gaming benchmarks show a ~9% improvement; I imagine the difference is due to improved memory latency although again this is just speculation. Of course for the gaming benchmarks we have no idea how many cores are in use so it could be literally anywhere between 3.9 and 4.3 GHz for each individual game.

Then according to that Ryzen 2### has lower gaming IPC than Ryzen 1### given the gaming performance difference is 3.4% with that 5.5% higher clock rate.

That would be pretty bad for AMD, they will get slaughtered for releasing a lower IPC CPU.

If Ryzen 2 is boosting then make that IPC deficit to Ryzen one near 10%

HeX · 26 Mar 2018 at 14:28

DragonQ said:
The 12 nm node certainly provides no power usage improvements. As expected it's an overclocked refresh, a la Radeon Rx 5xx series. I wouldn't expect IPC changes except for supporting faster RAM but speculating on IPC based on those graphs is foolish since no clock references are given.

Ryzen on 14nm LPP hit a voltage wall past ~3.8GHz. 12nm refresh wasn't about lowering power usage, it's aimed at moving that voltage wall and achieving higher clocks, which it appears to have done.

DragonQ · 26 Mar 2018 at 14:29

humbug said:
Then according to that Ryzen 2### has lower gaming IPC than Ryzen 1### given the gaming performance difference is 3.4% with that 5.5% higher clock rate.

That would be pretty bad for AMD, they will get slaughtered for releasing a lower performance CPU.

That's not necessarily true, clock scaling isn't linear and it's not far off the expected figure. There's a decent chance the 1800X is using XFR I guess but this is why I said speculating on IPC from these graphs is largely pointless. There could also be GPU limited titles in their benchmark suite dragging the numbers down, who knows since we have no settings for them. We need to see some hard numbers for individual benchmarks and games with known settings.

HeX · 26 Mar 2018 at 14:32

1800X should certainly be using XFR, only way it wouldn't is if they'd used some truly terrible cooler on it, or benchmarked it next to a radiator on full blast.

We know the 2700X isn't using XFR2 because it's not supported on non 4XX series mobo's. Whether it's defaulted to using some basic XFR mode we can't say.

humbug · 26 Mar 2018 at 14:36

NinjaCool said:
Videocardz has some info up on the site.
https://videocardz.com/75504/cpc-hardware-amd-ryzen-7-2700x-review-goes-online

Something is starting to smell a bit funny about this, if the gaming performance and the chart below are correct the 2700X does actually have a 4.1% IPC deficit, according to that chart the 2700X is running 'at least' 7.5% higher clock speed and yet is only 3.4% faster.

I thought tightening up the Inter CCX latency was supposed to make the gaming performance better, not worse.

Yeah i think i'm going to write this one off and file it under the same category all other Ryzen 2 leaks to date, fake click bait trash.

DragonQ · 26 Mar 2018 at 14:39

Why are you only considering single-core turbo speeds? Not one of the applications in their benchmark suite will be using just one core. Also, doesn't XFR only affect up to two cores? So that might not be relevant either.

humbug · 26 Mar 2018 at 14:45

DragonQ said:
Why are you only considering single-core turbo speeds? Not one of the applications in their benchmark suite will be using just one core. Also, doesn't XFR only affect up to two cores? So that might not be relevant either.

According to them the 4 core boost on the 2700X is 4.05Ghz, this vs 3.7 on the 1800X, that's a difference of 9.5% and yet they say the gaming performance is only 3.4% higher.

Thats an IPC deficit of 6.1%

I think its yet another fake leak.

sideways14a · 26 Mar 2018 at 14:52

The 1800 likely is boosting with xfr as well, those will be old figures they must have got for previous reviews.
The 2xxx chips are on a naff board so prob no xfr and god knows what effect on boost.

So.. we know... nothing