• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

*** AMD "Zen" thread (inc AM4/APU discussion) ***

humbug said:
Thats a better response... edited after my post but a response none the less.

The article you refer to is over 3 years old, the image AMD used is one they made themselves, its one of a Zen CPU with a Zen logo on it, given that the article is more than 3 years old it is impossible that Toms use that same image as it did not exist over 3 years ago, there for it is different and takes an unknown different amount of time to render, there fore those times cannot be used to compare with AMD times on todays event.

I'm also pretty sure AMD used the latest Blender, which again did not exist 3 years ago.
Click to expand...

Arrrggg can we please stop arguing about this!? The topic has died and been re-incarnated as the potato that your argument is resembling.

Tom's test looked at the comparable performance of AMD vs Intel when they tested it.

AMD's new test compared a different two chips carrying out a completely different test.

Can you still use this information to try and correlate performance? Yes, it will be flawed but still should be indicative of performance.
 
Yes ^^^ I too want to close on this, so to do so...

Toms times 3 years ago using a different image cannot be used to judge AMD results from today because Toms could be anything from 1% to Infinitely harder or easier to render than AMD's results from today, so if you know Toms takes 1 minute and AMD's event today takes 2 minutes it is utterly meaningless as we do not know how hard / easy and therefore fast or slow Toms image was, again, its like benchmarking one game and then a different game and comparing the FPS expecting them to be the same, of course not. :)
 
Last edited:
Troezar said:
If anything like close that would be a massive shake up...
Click to expand...

Well, we didn't get to see framerate counters directly but they had the SR7 3.4Ghz running against a 6900K in various tests and it was happily on par/maybe a nose ahead (was 54 secs on a blender test vs 59 for the 6900k). The video's all over the place so have a butchers.
 
No, but here is the important thing. Because they are being tested against each other we know what they are comparable to.

Say we go to the pool today and swim ten lengths each and you beat me by one length. We would know that you are about 10% faster at swimming on that day than me.

Next week you go swimming with Martin in a different pool and he beats you by 3 lengths... so he is about 30% Faster than you.

Can we be reasonably sure that Martin is faster than me? Yes. I may have gone swimming every day and done a heap of roids, but we can reasonably guess.

You obviously don't think that this is a reasonable model - why not?
 
Beren said:
No, but here is the important thing. Because they are being tested against each other we know what they are comparable to.

Say we go to the pool today and swim ten lengths each and you beat me by one length. We would know that you are about 10% faster at swimming on that day than me.

Next week you go swimming with Martin in a different pool and he beats you by 3 lengths... so he is about 30% Faster than you.

Can we be reasonably sure that Martin is faster than me? Yes. I may have gone swimming every day and done a heap of roids, but we can reasonably guess.

You obviously don't think that this is a reasonable model - why not?
Click to expand...

Because they are using different images, different images = different levels of difficulty = different times to complete.

To use your analogy: one day you are swimming in treacle, me and Martini in water, on another day Martini is swimming in treacle while you and me are in water..........
 
humbug said:
Where is this link you speak of? a simple question would be did Toms use AMD's Image from this event? if they didn't the comparison is flawed.
Click to expand...

Basically, the tests they did on the presentation are available on the site so you can compare it to your own kit.
Download it, run it, see what you get.
You can, of course, be suspicious about the downloaded test vs what they ran "live" but... that would be open and blatantly lying in a massive way. They need to get this one right so I don't think they'd risk it.
 
Mercutio said:
Basically, the tests they did on the presentation are available on the site so you can compare it to your own kit.
Download it, run it, see what you get.
You can, of course, be suspicious about the downloaded test vs what they ran "live" but... that would be open and blatantly lying in a massive way. They need to get this one right so I don't think they'd risk it.
Click to expand...

Yes, now, and that's valid, Toms 3 years ago did not have access to that image, its one AMD made, it has a Zen CPU picture on it for #### sake. :p

The version of Blender 3 years ago was also different to the one used today, as has been pointed out by another member in here older versions perform differently.

I find myself repeating that too much now ^^^^^
 
humbug said:
Yes, now, and that's valid, Toms 3 years ago did not have access to that image, its one AMD made, it has a Zen CPU picture on it for #### sake. :p

The version of Blender 3 years ago was also different to the one used today, as has been pointed out by another member in here older versions perform differently.

I find myself repeating that too much now ^^^^^
Click to expand...

Nah, I've got no issue with your point.

It might as well have been the time taken to render a 4 polygon potato vs a VR quality inside and out of St Pauls or something. They're clearly very different tests.

I'm suggesting only that folks can go grab what they used for the testing in the demo today and see what their own machine runs it like, compare from that. I wasn't getting involved in your argument at all :D

Download the demo, run it, record time. It's ONE test so we can only use it as a vague milestick but it still looks rather promising. It's high end haswell-e for the masses at i7 prices as things look at the moment.

I jumped on Sandy bridge early enough to have the B3 sata flawed chipset issue (though... erm.. Bolton, rhymes with "ham" screwed up and refunded my entire order when I returned the faulty P67 board so my entire sandy platform cost me a Z68 motherboard :D ) so I might wait a TINY bit this time, see what comes in March and where everything sits benchmarks wise but, as things sit I'm going AMD in around March/April next year.
 
Last edited:
Mercutio said:
Nah, I've got no issue with your point.

It might as well have been the time taken to render a 4 polygon potato vs a VR quality inside and out of St Pauls or something. They're clearly very different tests.

I'm suggesting only that folks can go grab what they used for the testing in the demo today and see what their own machine runs it like, compare from that.
Click to expand...

Thanks. and lol :)
 
In blender you can try clicking on the camera Icon on the right side, expanding Sampling, and changing the amount of Render samples to 100 instead of 200.

makes the image render a lot faster on my system when set to 100

please bear in mind we dont know what settings amd used in todays demo.
 
So much for the AM4 X370 motherboards being on show, wccftech are full off bull ****, lol >> http://wccftech.com/amd-zen-am4-x370-preview-new-hrozion/

Just out of curiosity, I installed Blender using the BMW Viewport demo with version 2.64a from October 2012 and got 13.10 seconds, I then ran version 2.78 and got 9.35 seconds, 40% improvement using the latest version pmsl. :confused:

Anyway, it was good to see the performance shown for the first time with AMD's Ryzen, I'll be more interested at the time when review sites show their results and we can get some conclusive results.
 
Last edited:
remember this:

http://wccftech.com/official-amd-bu...k-slides-leaked-performance-pricing-detailed/

If AMD really have a game changer, they had to do so much more to prove as far as I am concerned....a couple of cherry picked benchmarks and game demonstrations that would not be limited by the CPU........and with no pricing or release dates....


Not convinced at all.


Edit: http://cdn.wccftech.com/wp-content/uploads/2011/09/amdfxpressdeck_12a_dh_fx57.jpg

** Do Not Hotlink images **

(See! FX 8150 matches and even beats the i72600K in popular benchmarks!) :|
 
Last edited by a moderator:
I searched this thread and no one posted it.


https://www.pcper.com/reviews/Proce...Zen-CPU-Reveals-34-GHz-clock-and-more-details

PCPer recorded video on 8 December 2016 with much better quality footage to see full details zoomed in showed Handbrake performance and Blender performance with power consumption demo. With Blender app opened saw Ryzen consumed 93W and Intel 6900K consumed 106.5W, when it rendered Ryzen CPU workload saw Ryzen consumed 188.2W and Intel 6900K consumed 191.5W at full load.
 
Last edited:
Mercutio said:
Basically, the tests they did on the presentation are available on the site so you can compare it to your own kit.
Download it, run it, see what you get.
You can, of course, be suspicious about the downloaded test vs what they ran "live" but... that would be open and blatantly lying in a massive way. They need to get this one right so I don't think they'd risk it.
Click to expand...

Look at JonSmiths post, using an older version of blender is faster with AMDs fx 83. a lot faster
 
You must log in or register to reply here.
Back
Top Bottom