AMD Ryzen benchmarks against 7700K and 6900K

[Space Reptile]

cinebench and blender are similar as they both render a picture and calculate and a score is calculated from that , difference is that CB has locked settings and always renders the same thing , and as we saw the 6900 and Ryzen where head to head when rendering the same thing in blender (if im wrong , please correct me , I assume blender and CB are similar)

 

this is either a fake leak, or whatever hardware the person has is severly bottlenecked by either bad drivers or lackluster support from the hardware they used (pre release silicone?)

 

but then again , its not official and we all shud only trust official benches when the hardware hits

[Space Reptile]

4 hours ago, RKRiley said:

Without an actual trusted source, no matter who beats who, noone would believe it.

well I mean he would , his sources are no better

[patrickjp93]

9 minutes ago, Space Reptile said:

cinebench and blender are similar as they both render a picture and calculate and a score is calculated from that , difference is that CB has locked settings and always renders the same thing , and as we saw the 6900 and Ryzen where head to head when rendering the same thing in blender (if im wrong , please correct me , I assume blender and CB are similar)

 

this is either a fake leak, or whatever hardware the person has is severly bottlenecked by either bad drivers or lackluster support from the hardware they used (pre release silicone?)

 

but then again , its not official and we all shud only trust official benches when the hardware hits

Or, since blender is OpenSource, AMD chose an image that played nicely with its L2 cache size but not Intel's, or, even better, MODIFIED SPECIAL SOURCE CODE FOR THE RYZEN TEST!

 

AMD faked the tests. Are you really so surprised?

6 minutes ago, Space Reptile said:

well I mean he would , his sources are no better

They're always better. They're mine.

[WereCat]

5 hours ago, Ryan_Vickers said:

 

it's a photo so it's totally legit /s

Wait... is this the new Zen dual core? Intel R.I.P.!!!

[Prysin]

31 minutes ago, patrickjp93 said:

DB has claimed no such thing.

 

You can have a 40% gain in one area and a much smaller gain in another. If Intel's fmull went from 2 cycles per instruction to 1, it would be a 100% increase. The amount most people would actually feel because of how instructions tend to be strung together is probably 10% better or less!

https://mobile.twitter.com/Dresdenboy/status/811013654011383808

 

ahem, he didnt?

He suggests that it is a E4 stepping, whilst the latest we saw was a F3 stepping...

[leadeater]

Some crude math actually suggests that score (1188) isn't too far wrong, I would have expected higher and I'm still very suspect. If you take AMD's claim of 40% IPC increase and use AMD's best single thread CB score of 110 x 1.4 x 8 = 1232. Add a little extra for 'HT' and not yet finalized firmware and clocks and I can see Zen getting very similar CB scores to the 6900K.

 

But we all know how useless CB scores actually are though right??....... Ignoring all those real graphical artists out there that likely aren't in this thread at all.

[patrickjp93]

12 minutes ago, Prysin said:

https://mobile.twitter.com/Dresdenboy/status/811013654011383808

 

ahem, he didnt?

He suggests that it is a E4 stepping, whilst the latest we saw was a F3 stepping...

And it's a wild guess with no source, not any actual claim of his. The boost clock also matches the retail product so...

[Prysin]

16 minutes ago, patrickjp93 said:

And it's a wild guess with no source, not any actual claim of his. The boost clock also matches the retail product so...

you're so precious sometimes. 

 

On 28/11/2016 at 9:34 AM, patrickjp93 said:

The risk is much greater than the reward imho, and both Zen and Vega would have to exceed market sales expectations enormously to cover the gap in market cap and financial health that reflect the real price of the stock (roughly $4.98 coming from a former financial stock broker).

 

He's a former Global Foundries engineer you twit.

 

No, he has NEVER been wrong.

 

 

 

On 28/11/2016 at 0:11 AM, patrickjp93 said:

Dresdenboy has not been wrong once in his six years of sourcing AMD leaks.

 

Science is about proving, not experimenting. The scientific method only shows validity of correlation hypotheses. It doesn't prove anything, not on its own.

 

 

your move patrick. Also, your words.

[patrickjp93]

5 minutes ago, Prysin said:

you're so precious sometimes. 

 

 

 

 

your move patrick. Also, your words.

He's never been wrong when he's leaked AMD info. This is not one such time. He's speculating on someone else's incomplete leak. That's not the same thing.

 

You're quite precious when you try to go out of your depth, especially when you conflate things you shouldn't.

[Prysin]

5 minutes ago, patrickjp93 said:

He's never been wrong when he's leaked AMD info. This is not one such time. He's speculating on someone else's incomplete leak. That's not the same thing.

 

You're quite precious when you try to go out of your depth, especially when you conflate things you shouldn't.

thing is, If he has as much knowledge about AMD products as you claim he has. His hypothesis of this leak being a older ES should be within realm of 90%+ true. He would know what the latest ES should fetch. He would also know the rough overhead from incomplete optimization in CB...

 

So either he isnt nearly as "in the know" as you claim him to be. Or your narrative is all wrong.

[patrickjp93]

6 minutes ago, Prysin said:

thing is, If he has as much knowledge about AMD products as you claim he has. His hypothesis of this leak being a older ES should be within realm of 90%+ true. He would know what the latest ES should fetch. He would also know the rough overhead from incomplete optimization in CB...

 

So either he isnt nearly as "in the know" as you claim him to be. Or your narrative is all wrong.

Based on what?

 

IPC does not improve from stepping to stepping. What has you people convinced it does?

 

My narrative is correct, he's in the know, but it's still a wild guess. He's also not making a definitive claim. He says that's his best guess based on AMD's blender score, but that score's a manipulated pile of crap, so that doesn't mean much.

[LAwLz]

1 hour ago, patrickjp93 said:

If Intel's fmull went from 2 cycles per instruction to 1, it would be a 100% increase. The amount most people would actually feel because of how instructions tend to be strung together is probably 10% better or less!

What's funny is that when I said this exact thing about AVX, that just because one instruction gets faster, the entire program won't by the same degree, you disagreed.

[patrickjp93]

2 minutes ago, LAwLz said:

What's funny is that when I said this exact thing about AVX, that just because one instruction gets faster, the entire program won't by the same degree, you disagreed.

I most certainly did not. Feel free to quote me. It's going to be so satisfying seeing a charlatan like you leave when Zen hits retail and those disappointing benchmarks come in.

[leadeater]

Let's all take some time to contemplate this cute kitten and feel it's calming effects, remember to keep debates civil.

 

[patrickjp93]

3 minutes ago, leadeater said:

Let's all take some time to contemplate this cute kitten and feel its calming effects, remember to keep debates civil.

Civility doesn't require that I refrain from laughing at others' expense, especially Lawlz'.

 

Also, ftfy.

[LAwLz]

10 minutes ago, patrickjp93 said:

I most certainly did not. Feel free to quote me. It's going to be so satisfying seeing a charlatan like you leave when Zen hits retail and those disappointing benchmarks come in.

Sorry, I can't find the exact post but here is one of my old posts about it. You left the conversation without answering me though.

 

On 2016-04-02 at 0:19 PM, LAwLz said:

What the hell are you talking about? Are you going to prove to me that a video encoder uses AVX? I never said they didn't. In fact, I know which article you are talking about and the article was about x264... x264 does use AVX. The developer I talked to said that using AVX2 did almost double the performance in some parts of the program, but those parts only make up a fairly small part of the overall runtime so the performance gains were not anywhere near 100%.

 

Please bear in mind that I am not making any claims or picking side whether or not Zen will be good or bad. I am just saying that I completely agree with your claim in this thread that an improvement to one instruction does not mean an overall program will gain the same % performance increase.

It feels nice that we agree for once.

[patrickjp93]

3 minutes ago, LAwLz said:

Sorry, I can't find the exact post but here is one of my old posts about it. You left the conversation without answering me though.

 

 

Please bear in mind that I am not making any claims or picking side whether or not Zen will be good or bad. I am just saying that I completely agree with your claim in this thread that an improvement to one instruction does not mean an overall program will gain the same % performance increase.

And so your problem with me is what?...

[patrickjp93]

10 minutes ago, LAwLz said:

It feels nice that I agree with you for once.

ftfy.

[Master Disaster]

AMD Thread created on LTT...

 

[patrickjp93]

1 minute ago, Master Disaster said:

AMD Thread created on LTT...

-Non-Dank Meme-

 

[That Norwegian Guy]

TFW people think Ryzen is going to compete against 6900K just because that's what they compared it to on the show floor.

 

This is a consumer i7 competitor, not an enthusiast i7 competitor.

[porina]

5 minutes ago, That Norwegian Guy said:

TFW people think Ryzen is going to compete against 6900K just because that's what they compared it to on the show floor.

 

This is a consumer i7 competitor, not an enthusiast i7 competitor.

Given they're going to use the same platform unlike Intel, they have the ability to blur the lines and it wont be so clear cut.

[AxelRantila]

I am waiting for proper independent benchmarks. So much rumors going around.

[Prysin]

36 minutes ago, That Norwegian Guy said:

TFW people think Ryzen is going to compete against 6900K just because that's what they compared it to on the show floor.

 

This is a consumer i7 competitor, not an enthusiast i7 competitor.

RyZen should compete with Haswell-E. Even Broadwell-E is a bit of a stetch

[LAwLz]

55 minutes ago, patrickjp93 said:

ftfy.

If I agreed with you more often, then we'd both be wrong so many times.

 

 

7 minutes ago, Prysin said:

RyZen should compete with Haswell-E. Even Broadwell-E is a bit of a stetch

Like always, it's better to set your expectations low and be pleasantly surprised than to put your expectations too high and get disappointed.

It seems like some people are expecting 6900K performance for 6700K price and those will most certainly be disappointed.