****Blender Ryzen render battle benchmark****

[implosiveturnip]

I know most people will probably say the latter is of no importance, but there's a good reason they've not discussed overclocking on both core and memory. i.e it's non existent.

I'm thinking the good reason being that overclocking is a very niche market. Most people don't really care for it; the gamers that they are clearly gunning for with this release is a much bigger market, and the majority of them will just want a fast, affordable chip out of the box, without any tweaking.

I'm going to speculate that we might see respectable results to be honest, considering that the chip's base is now going to be 3.4ghz, which is up from the first leaked samples at 2.8ghz, and the first blender test they did which was at 3.0ghz (iirc).

At least that's what I'm hoping. Historically AMD's CPU's overclock quite well, but their memory controllers are a bit weaker than Intel's, which might hold things back on the memory side. The lower TDP of this chip and smaller process node should help to yield good results, but until we actually get some details, it is all just speculation.

Hurry up dammit

 

[Rainmaker]

My stock FX8350 (with 16GB 1600 RAM) managed 2m 10.38s at 150 samples on Windows 10 Pro x64. Zen will be a nice step up either way. I've told my wife to belay the Xmas presents until January...

 

[Silent_Scone]

I'm thinking the good reason being that overclocking is a very niche market. Most people don't really care for it; the gamers that they are clearly gunning for with this release is a much bigger market, and the majority of them will just want a fast, affordable chip out of the box, without any tweaking.

I'm going to speculate that we might see respectable results to be honest, considering that the chip's base is now going to be 3.4ghz, which is up from the first leaked samples at 2.8ghz, and the first blender test they did which was at 3.0ghz (iirc).

At least that's what I'm hoping. Historically AMD's CPU's overclock quite well, but their memory controllers are a bit weaker than Intel's, which might hold things back on the memory side. The lower TDP of this chip and smaller process node should help to yield good results, but until we actually get some details, it is all just speculation.

Hurry up dammit

Are you saying gamers don't overclock their CPU? Not that it matters if you are, as that has nothing to do with why Zen has very little range.

 

[humbug]

Getting 3.4Ghz from the 8 core at <95 watts i very much doubt Zen has no overclocking headroom, that is 200Mhz (7%) higher base clock and with less power than Intel 8 core so this doesn't look like something that is at anything like its limit.

Add to that it does have boost (self overclocking) AMD just ain't saying by how much, probably because Intel 7### series is just a pre-overclocked 6### series and they don't want to give away what its true performance will end up at.

 

[CAT-THE-FIFTH]

Getting 3.4Ghz from the 8 core at <95 watts i very much doubt Zen has no overclocking headroom, that is 200Mhz (7%) higher base clock and with less power than Intel 8 core so this doesn't look like something that is at anything like its limit.

Add to that it does have boost (self overclocking) AMD just ain't saying by how much, probably because Intel 7### series is just a pre-overclocked 6### series and they don't want to give away what its true performance will end up at.

Check the Zen APU thread - it looks the hype tram might have delivered for once!!

The ES tested by the French magazine seems almost clock to clock similar to the Core i7 6900K even in games.

 

[humbug]

Check the Zen APU thread - it looks the hype tram might have delivered for once!!

The ES tested by the French magazine seems almost clock to clock similar to the Core i7 6900K even in games.

Yeah i did see that , the only two hopes that are left really are that; A, 'its no less expensive than Intel' and / or; B, '3.4Ghz is all it will clock to'

#StrawHopes

 

[MichaelM]

No idea how to change or check sample quality so I just ran the two files posted.

5930K @ 4.75GHz
150 samples? = 36.77 seconds
200 samples? = 48.67 seconds

 

[implosiveturnip]

Are you saying gamers don't overclock their CPU? Not that it matters if you are, as that has nothing to do with why Zen has very little range.

I was saying that they are trying to appeal to a wider market on the limited exposure they have given it, which is why they may have skipped over mentioning overclocking, as only a small group of people are interested in it compared to the wider market.

I really do hope it clocks well, hower as optimistic as I am about Zen, and what we've seen it doesn't look too shabby at all. However, considering how many times AMD has failed to deliver in recent years,you might be completely right about the lack of info here and it's a total lemon. I'm banking on nothing until we have independant reviews out though.

 

[siriq111]

AMD FX [email protected]
150 sample 01:07:03
I think i will keep this rig for awhile . No need to upgrade

[/IMG]

 

[siriq111]

I think i will keep my rig. Zen with 16 threads(stock), does 36 sec and FX [email protected], does 67 sec. Despite the fact FX is on OC, still holding up , compare to stock Zen(8/16 threads). I would say, Zen 4/8 will be around 20% faster in most scenarios compare to my rig. Better to just save some money for myself, for REV 2.0 Zen. Let me know about your odds

 

[Rainmaker]

I think i will keep my rig. Zen with 16 threads(stock), does 36 sec and FX [email protected], does 67 sec. Despite the fact FX is on OC, still holding up , compare to stock Zen(8/16 threads). I would say, Zen 4/8 will be around 20% faster in most scenarios compare to my rig. Better to just save some money for myself, for REV 2.0 Zen. Let me know about your odds

How do you come to that conclusion? By your own testing, Zen is almost twice as fast as your overclocked FX8350 (and 4x faster than my stock one) - certainly in Blender. That's not really a small upgrade.

BTW, OP, according to the small print at the end of AMD's Blender video (YouTube) they used Blender v2.77a not 2.78a. The former is quite a bit faster than the newer one.

 

[crazyDAJT]

X5650 @ 4.4GHz

200 (default) - 1:09.60
150 - 52.07
150 - 35.50

Not to bad for a over the hill system

so approximately 40-45% slower than Ryzen, not surprising considering it has an extra 2 cores and 4 threads

 

[siriq111]

150 samples FX [email protected]

100 samples:

 

[xs2man]

How do you come to that conclusion? By your own testing, Zen is almost twice as fast as your overclocked FX8350 (and 4x faster than my stock one) - certainly in Blender. That's not really a small upgrade.

Sure, it's almost twice as fast, but at what multiplier of cost? The Zen prices aint been released yet, but what do we expect for the 8c/16t Zen? £600+ I would imagine, compared to circa £150 for a FX8350. So half the time for 4 times the price.

Now if a 4c/8T comes in at around the £200 mark, with that 20% increase, then yeah, fair enough. Probably worth an extra 25% for 20% performance and all the other benefits that go along with the new platform. Otherwise, I can see his point.

 

[daveyb]

x5650 4.0ghz

200 samples 1:15.78s
150 samples 58.43s
100 samples 38.57s

 

[JonSmith]

Hey guys, if you recall that we had some suspicions about the Blender performance of Piledriver not being in line with other benchmarks, the Stilt recompiled it with SSE switches on and got significantly better performance (literally halves the render times) whilst Intel chips do not see such a big improvement.

Some of this is worth a read here:

https://forums.anandtech.com/thread...ark-confusion-run-150-samples.2494600/page-10

 

[Genoma]

6C/12T Xeon 5650 @ 4.0GHz here too :

150 = 58.93
100 = 39.32

not too bad for a £60 CPU on my 7 year old X58 mobo

 

[Silent_Scone]

6850K @ 4.6Ghz / 3.6Ghz Uncore (6C 12T)

150=33.10

Spoiler

 

[pete910]

^^^ Spoiler tags anyone

 

[Silent_Scone]

heh, sorry I forgot I was posting on pauper net.

It scales down for me in Chrome so didn't notice!