Athlon X4 845 ExcavatorIPC benchmarks

[Erenhardt]

What about them? Nobody is going to consider buying an I3 for rendering work.

Nobody is rendering in CB. Single thread mode in particular!

 

[TheELF]

CB is a bench for maxon's cinema 4d which is a 3d rendering suit.

It's being used as a benchmark for ages because it's one of the few professional softwares that can use 100% of each core available.
Want proof? Run this bench with twice the threads per cores and you will get the same result or very slightly below,run 7z or x264 bench with twice the threads per core and you will see considerably higher scores,on intel at least never tried it on amds,because those are public domain,amateur softwares,they are great tools but they lack the programming skills of an established company.
And that's why people think that CB is biased when all it is is efficient(optimized,what ever) .

 

[Abwx]

That s not the point wich is that there are people here who consider that if the i3 is better in CB ST and the FX better in CB MT then the result is a tie....

 

[Erenhardt]

CB is a bench for maxon's cinema 4d which is a 3d rendering suit.

Using inferior software (cinema 4d) to measure which CPU is superior seems like not the wise idea.

because those are public domain,amateur softwares,they are great tools but they lack the programming skills of an established company.
.

Reads like "it is easier to bribe the company than to bribe some pesky enthusiast".

But hey, your opinion is just as valid as mine and everones else. :thumbsup:

 

[TheELF]

Using inferior software (cinema 4d) to measure which CPU is superior seems like not the wise idea.



Reads like "it is easier to bribe the company than to bribe some pesky enthusiast".

But hey, your opinion is just as valid as mine and everones else. :thumbsup:

No matter how crappy cinema 4d is it can make better use of cores.
I told you how to test this on your own.
But hey,it's easier just to follow the blame train believing the conspiracy nuts than to actually spend some time running some tests.

 

[TheELF]

That s not the point wich is that there are people here who consider that if the i3 is better in CB ST and the FX better in CB MT then the result is a tie....

...for rendering or DC scenarios.
And you are right it's not a tie and the FX are in a big lead.

But there are also people here who consider that if the FX is better in CB MT and the i3 is better in CB ST then the result is a tie.... for gaming

 

[ShintaiDK]

Yeah, which is why people are saying Sandy Bridge IPC for Zen. Something like an 8 core Zen with a 3-3.2 Ghz base really would not be that bad for games given that games are starting to use 6 threads pretty effectively. Zen's appeal beyond gamers is really limited though.

Yep, we already see datacenters selling out of old hardware that Zen may actually match if they are lucky. SB performance, IB performance/watt perhaps.

And we are back to a gaming CPU that requires OC to be relevant.

 

[dark zero]

...for rendering or DC scenarios.
And you are right it's not a tie and the FX are in a big lead.

But there are also people here who consider that if the FX is better in CB MT and the i3 is better in CB ST then the result is a tie.... for gaming

In fact in Home and some semi pro apps, where ST rules, the i3 wins, however on MT apps like vídeo renderer or virtual machines, the FX is still useful.

However that excavator... Is a massive nerf compared to Vishera

 

[The Stilt]

For modern, relevant video codecs (HEVC, VP9) current AMD CPUs are basically useless. Any currently available AMD CPU going against Intel CPUs in X265 is a pure blood bath.

 

[USER8000]

For modern, relevant video codecs (HEVC, VP9) current AMD CPUs are basically useless. Any currently available AMD CPU going against Intel CPUs in X265 is a pure blood bath.

It appears so,at least for X265:

http://x265.ru/en/x265-hd-benchmark/

x265 HD Benchmark is a benchmark that measures how fast your computer can encode HD (1080p) video in the new H.265 / HEVC format.

On their website the x265 HD Benchmark people have ranked CPUs,and an FX8300 CPU is similar clock to clock to an Ivy Bridge Core i7. It looks like you need a 4.8GHZ FX8300 series CPU to match something like a Core i7 4790K,running at 4.0GHZ to 4.4GHZ it appears.

I suspect a Core i7 6700 is out of bounds to any current AMD CPU.

Thats the problem with the FX8300 series CPUs being a 2012 design launched just after Ivy Bridge.

 

[superstition]

It looks like you need a 4.8GHZ FX8300 series CPU to match something like a Core i7 4790K,running at 4.0GHZ to 4.4GHZ it appears.

Given that an 8320E costs $90 at Micro Center and can be overclocked to that speed I'm not sure it's really such a bloodbath.

However, it will use a lot of power and need a robust system.

But, the performance per dollar seems reasonable enough, especially if you get the $40 off a board deal. If you don't have a Micro Center near, though...

 

[superstition]

There is no "architecture bias" in Cinebench. Cinebench R11.5 uses MMX, SSE and SSE2 instructions and R15 uses MMX, SSE, SSE2 and SSE3.

AMD CPUs have inferior FPU performance compared to Intel and that's the only reason why they perform bad in Cinebench.

They perform just as bad or even worse in open-source FP heavy application / benchmarks (C-Ray, Rodinia Euler CFD, X265, VP9, etc). In X265 Skylake is around 90% faster than Excavator due the comprehensive AVX2 support. Skylake gets around 30% boost from AVX2 alone, whereas Excavator gets additional performance penalty of 2-5%.

Pov-Ray is integer, X264 is integer, 7-Zip (LZMA) is integer. That's the reason why 15h can keep up even remotely in these tests.

You are discussing the bias, then, by showing that a dichotomy exists.

Given the dichotomy it is necessary to have two benchmarks at least and ideally a third (a blend of integer and FP) to avoid a biased picture.

Imagine if the only benchmarks people used to compare were pure integer. They would have no idea that FX has a FP weakness.

 

[Dresdenboy]

Very thanks, dresdenboy. Looks like I misunderstand, the x64 version of CPUZ is executing SIMD instead of pure integer, and using VC++. This is still a bench which is fair enough between different vendors and architecture.
I wonder if x32 version which use both x86 & x87 result is any differ than x64. (x32 ver is a kind of bench similar to superpi though......)

I'm not sure, but when I looked at the code, I think I saw scalar SSE ops in te 64b version. The code also used integer division a lot. Plotting the CPU-Z score/GHz/core did show some nice generational increases between uarchs.

 

[JDG1980]

Or lets face it 40% could be best case for a cherry picked benchmark.

I'm assuming that AMD is actually interested in competing. Lisa Su has said, on an earnings conference call, that Zen is on track for "greater than 40% IPC uplift". This is a statement to stockholders, where it would be illegal to flat-out lie. Putting together the existing public statements and filtering them with common sense and what we know about the Zen architecture so far from Dresdenboy's analysis and others, it seems reasonable that we'll be seeing more than 40% IPC improvements in FPU-heavy applications, and less in integer-heavy applications where BD already does acceptably well.

AMD needs at least Sandy Bridge-level IPC to compete, and they know it. Creating a hype train and letting down customers again like they did with Bulldozer is a ticket to bankruptcy, and they know that too. They are betting the company on Zen, and if it's not at least reasonably competitive, they're finished.

 

[Arachnotronic]

I'm assuming that AMD is actually interested in competing. Lisa Su has said, on an earnings conference call, that Zen is on track for "greater than 40% IPC uplift". This is a statement to stockholders, where it would be illegal to flat-out lie. Putting together the existing public statements and filtering them with common sense and what we know about the Zen architecture so far from Dresdenboy's analysis and others, it seems reasonable that we'll be seeing more than 40% IPC improvements in FPU-heavy applications, and less in integer-heavy applications where BD already does acceptably well.

AMD needs at least Sandy Bridge-level IPC to compete, and they know it. Creating a hype train and letting down customers again like they did with Bulldozer is a ticket to bankruptcy, and they know that too. They are betting the company on Zen, and if it's not at least reasonably competitive, they're finished.

She said "greater than 40% IPC uplift" in reference to their prior generation server architecture, which was Piledriver.

Interesting how subtle the manipulation on AMD's part is.

 

[JDG1980]

Yep, we already see datacenters selling out of old hardware that Zen may actually match if they are lucky.

Those LGA 2011 CPUs have no warranty and require a $300 motherboard, which makes them a little less impressive of a deal than they look at first. They also run on an outdated platform (e.g. no native USB 3.0 support), though admittedly not as outdated as AMD's FX chipsets.

SB performance, IB performance/watt perhaps.

And we are back to a gaming CPU that requires OC to be relevant.

IPC will probably be pretty close to Sandy Bridge. Perf/watt should be considerably better than Ivy Bridge-E and Haswell-E, due to the use of the 14nm LPP FinFET process. Remember, those Intel chips were on 22nm. Broadwell-E will likely have better perf/watt than Zen, but I don't think the difference will be huge.

And we don't have any solid information yet on what clock speeds will be like. AMD has historically cranked them up as high as they reasonably can. The fewer cores, the higher you can clock in a given TDP, so the quad-core salvage parts will probably be the best bet for gamers who don't want to overclock.

 

[JDG1980]

For modern, relevant video codecs (HEVC, VP9) current AMD CPUs are basically useless. Any currently available AMD CPU going against Intel CPUs in X265 is a pure blood bath.

I'm not sure I would go as far as to call x264 irrelevant. Many mobile devices don't support H.265 or VP9 hardware decoding yet, and software decoding would be too slow and/or power-hungry. So you'll still see H.264 used as a lowest-common-denominator format for some time to come.

 

[nismotigerwvu]

Pov-Ray is integer, X264 is integer, 7-Zip (LZMA) is integer. That's the reason why 15h can keep up even remotely in these tests.

Now I knew there were some major differences in how H264 and H265 approached things, but I didn't know that H265 was floating point heavy. It's remarkable to me that newer codecs would swing the type of workload that drastically.

 

[USER8000]

She said "greater than 40% IPC uplift" in reference to their prior generation server architecture, which was Piledriver.

Interesting how subtle the manipulation on AMD's part is.

Actually not really any manipulation happening,if you look at both of the charts/graphs.

They said 40% over Exacavator and over 40% versus Piledriver.

Last time I checked that is factually correct,since Excavator is faster than Piledriver.

 

[DrMrLordX]

What about them? Nobody is going to consider buying an I3 for rendering work.

Nobody's going to buy an x4 845 for rendering work, either.

 

[TheELF]

Actually not really any manipulation happening,if you look at both of the charts/graphs.

They said 40% over Exacavator and over 40% versus Piledriver.

Last time I checked that is factually correct,since Excavator is faster than Piledriver.

Yeah it's very factual them not showing you any numbers(difference between bull and exc) on this graph so you can't figure out if they mean single or multi core,40% faster core or 40% when the whole CPU is working.

The suspected more pipelines plus smt plus targeted server market sure seems to point towards the second case.

 

[jpiniero]

I think if the IPC is close to sandy bridge the clock better be closer to 4GHz than 3GHz and 6+ threads to make it really interesting for gaming (with competitive pricing)

It almost has to be 6 and 8 cores at launch. The base and turbo clock speed can't be that low but it (well it should be) very competitive with stock Skylake i5's in games. Remember games are starting to use 6+ threads pretty effectively and not having HT really hurts. Then again it'll have to compete with the 2600K/3770K/4770K as well.

 

[ElFenix]

edit: nm this thread has completely jumped the shark.

 

[USER8000]

Yeah it's very factual them not showing you any numbers(difference between bull and exc) on this graph so you can't figure out if they mean single or multi core,40% faster core or 40% when the whole CPU is working.

The suspected more pipelines plus smt plus targeted server market sure seems to point towards the second case.

Maybe,next READ what the guy was saying instead of reacting first. He tried to say that AMD was changing its claims. Don't try to defend his error.

Only somebody wanting to change the claims themselves would be getting confused when one claim says 40% over Excavator and another says over 40% over Piledriver. Excavator has higher IPC over Piledriver. Last time I checked Steamroller had higher IPC than Piledriver. There is no grey area here.

Seriously,I know this is the interwebs,but come on??!! Really??!!

Plus any IPC claim could mean anything - if Intel says their core has X percentage IPC improvement,are you questioning what it is in?? Is it a core with HT disabled or enabled?? No different than the situation with Zen with HT,as like with Intel it says Core,not CPU.

It could be anything. We will need to wait closer to release to get a better clue of what they mean.

 

[BigDaveX]

And we don't have any solid information yet on what clock speeds will be like. AMD has historically cranked them up as high as they reasonably can.

The problem is, that quite often hasn't been enough. K10 fell well short of the clockspeeds that it needed in order to compete with the Core 2 line, and even K8's clocks were a little below the ideal range at launch (though in that case it didn't much matter, since nVidia dragged their feet on making a desktop chipset for it anyway).

Not saying that AMD are bound to fall short with Zen, mind, but it really is a complete shot in the dark as to whether or not it's going to clock high enough to compete with Intel at the high end, even if they don't pull the Emergency Edition trick again and yank some ~4.5GHz Skylake/Kabylake chips out of their backside to spoil the launch party.