Zen Engineering sample benchmarked by french tech magazine CanardPC

[RagnarokDel]

Just now, RexinOridle said:

Good post bud

But in a language that most people don't understand, at 0.5 megapixels, with random code name, and never heard of this guy. All I'm saying is if rumors were the pre-alpha, and AMD's blender benchmark was alpha; this is then beta.

Will wait for an actual benchmark.

https://en.wikipedia.org/wiki/Canard_PC

[Liltrekkie]

4 hours ago, Energycore said:

We have no way to know if the lower core count chips will be good or not, yet

 

I know, BUT hear me out. If the 4 core version has the same IPC as the chip in this test. Let's just assume for a minute it's a cut down version of 8 core ryzen. 

 

Working on that assumption, 4 cores at current 3.2 GHz speed would consume less power and produce less heat. Now, in theory depending on power targets and thermal targets, that would leave room to up the voltage and clock speeds. Both Intel and AMD release multiple SKUs of their chips (see the 8350 and 8370 for example) with minor clock speed jumps. 

 

So with that, it is not urological to assume that a 4 core version will exist with a higher clock speed. Now how much higher is the question. But, if it even similarly reaches Skylake .clock speeds then AMD will have produced a chip that does in fact compete with skylake.

 

Everyone here is complaining that the 8 core version of Zen does not compete with skylake, but.. Intels own 6 and 8 core chips don't compete with it either, they are both slower than skylake.. 

[AresKrieger]

4 hours ago, Energycore said:

They did Far Cry 4 which I think is single threaded

Not single threaded in fact I don't think it could run without an i5 or higher at launch

[Energycore]

2 minutes ago, AresKrieger said:

Not single threaded in fact I don't think it could run without an i5 or higher at launch

I think I should have said "a single-threaded performance test" since it can't use the full 8 cores and 16 threads.

[AresKrieger]

3 minutes ago, Energycore said:

I think I should have said "a single-threaded performance test" since it can't use the full 8 cores and 16 threads.

Most programs in general can't, and its not just due to lazy programmers some tasks can only be divided to a certain extent due to their serial nature, games have a lot of these style calculations once graphics are no longer a factor.

[Energycore]

1 minute ago, AresKrieger said:

Most programs in general can't, and its not just due to lazy programmers some tasks can only be divided to a certain extent due to their serial nature, games have a lot of these style calculations once graphics are no longer a factor.

I'm not saying anything about whether it's right or wrong that Far Cry 4 can only use so many theads. Just stating things as matter of fact. The games tests are more indicative of per-core performance on CPUs that have 4/8 threads or more.

[leadeater]

1 hour ago, patrickjp93 said:

You have to determine TDP and clocks that a maximal number of chips can fall within tolerance of to be able to sell them in mass. That's the nature of the beast.

True but on a high end SKU something like 30% in my eyes is nothing more than bad optimization or intentional limitation, for example if they ran it at higher and easily achievable clock rate it may be too close performance wise to the product above it making it much harder for the consumer to justify the extra cost.

 

We live in the age of high tolerance and control, 30% would fly directly in the face of that. Not that I believe people are getting 30% OC over what the new GPU boost can do, so the original point I was responding to and my own is basically not valid anyway.

 

We already know Nvidia intentionally limits their products, every Ti card is proof of that. It's not a bad thing at all, if you could buy the Ti at launch of the new architecture not 6+ months later. Rarely is there anything done to Ti cards other than unlocking CUDA cores that were already there and functional with very minor firmware improvements, something you can do by releasing firmware updates (the firmware optimization).

[MageTank]

1 minute ago, leadeater said:

True but on a high end SKU something like 30% in my eyes is nothing more than bad optimization or intentional limitation, for example if they ran it at higher and easily achievable clock rate it may be too close performance wise to the product above it making it much harder for the consumer to justify the extra cost.

 

We live in the age of high tolerance and control, 30% would fly directly in the face of that. Not that I believe people are getting 30% OC over what the new GPU boost can do, so the original point I was responding to and my own is basically not valid anyway.

 

We already know Nvidia intentionally limits their products, every Ti card is proof of that. It's not a bad thing at all, if you could buy the Ti at launch of the new architecture not 6+ months later. Rarely is there anything done to Ti cards other than unlocking CUDA cores that were already there and functional with very minor firmware improvements, something you can do by releasing firmware updates (the firmware optimization).

Intentional limitations are what I define the non-K CPU's as. I've tinkered long enough with the locked Skylake SKU's to say without a doubt that all of them had loads of potential in them. Buying random G4400's, 6600T's and 6400/6500's, all of which hit 4.5ghz with ease (some clocked as low as 2.7ghz base). I understand WHY it's done, as something has to be done with the cutdown SKU's, and I understand Intel's segmentation strategy. However, It still seems wasteful in a sense to weaken a product to sell it in another price bracket. Granted, the solution isn't to release a hundred variations of a SKU either.

 

Not gonna lie, I miss that 6600T. Was good fun tinkering with that chip.

Can tell how old this image is, by the fact that my ram isn't 3600 C14-14-14-28-2 in it, lol. That was when I was dumb and naive, trying to force CR1 to work, before eventually giving in and going with CR2. Luckily my bandwidth and latency are better now at CR2 with tighter tertiary's, so I'm no longer bummed out.

[DELTAprime]

These benches don't help me make a decision between a 7700k and Ryzen. Bring on the gaming benchmarks reviewers.

[Fetzie]

5 hours ago, Tocsin_786 said:

You should play Planet Coaster. Once you get 5k+ guests in the park, the cpu and ram usage maxes out and the frames drop to 5FPS. Same with Cities Skylines. Soon it will be a lot more games that arent sim builders that will benefit from more cores. 

Yeah, I have a Cities Skylines map that currently has all 12 threads of my 4.5 GHz 5820k at 70-80%. Finalising a new public transport line takes about 5-10 seconds. It was completely unplayable on my 2500k at 4.5 GHz at 150k less population (currently around 450k). The game also uses 11 GB of RAM.

[SCHISCHKA]

i just want to say how refreshing it is to see a benchmark in french. french speaking people are a brutally honest bunch

[leadeater]

2 hours ago, MageTank said:

Intentional limitations are what I define the non-K CPU's as. I've tinkered long enough with the locked Skylake SKU's to say without a doubt that all of them had loads of potential in them. Buying random G4400's, 6600T's and 6400/6500's, all of which hit 4.5ghz with ease (some clocked as low as 2.7ghz base). I understand WHY it's done, as something has to be done with the cutdown SKU's, and I understand Intel's segmentation strategy. However, It still seems wasteful in a sense to weaken a product to sell it in another price bracket. Granted, the solution isn't to release a hundred variations of a SKU either.

I actually completely agree with doing it for lower SKUs since they all use the same die, quite often ones that would pass a much higher grade but due to demand they need to be packaged lower. I just think it's less ok for higher end products, particularly if we look at the Nvidia 1070 vs 1080. They both use the same die yet the 1070 has a lower clock rate, why? It already has less CUDA cores enabled and GDDR5, which won't force you to run a lower core clock.

 

If your putting down large amounts of money on something you should expect the best out of it. If we look at Intel HEDT CPUs there should be functional/feature differences between all models and all run at the highest clock rate possible: 6 core with limited PCIe lanes (6800K), 6 core full PCIe lanes (6850K), 8 core (6900K) and finally a 10 core (6950X). With the exception of the i7 6850K running 200Mhz higher clocks all round than the i7 6800K I say bravo to Intel for actually giving a range of products that make sense and justify their cost increases. We could sit around arguing about the high cost of these products but Intel really has done a very customer first product lineup with these by offering real tangible benefits for your extra money.

[DELTAprime]

7 minutes ago, leadeater said:

If we look at Intel HDET CPUs

I'm sorry, what does HDET stand for, I'm seeing it a lot and Google can't figure it out for me?

[Fetzie]

3 minutes ago, DELTAprime said:

I'm sorry, what does HDET stand for, I'm seeing it a lot and Google can't figure it out for me?

Should be HEDT: "High End DeskTop"

[DELTAprime]

1 minute ago, Fetzie said:

Should be HEDT: "High End DeskTop"

Thanks. I was thinking it must be some type of technology like SSE. lol

[leadeater]

1 minute ago, DELTAprime said:

Thanks. I was thinking it must be some type of technology like SSE. lol

No I just durped and screwed up HEDT with HDET lol.

[leadeater]

Please, my AMD E2-1800 is the fastest laptop ever. It can run calculator, it does take a while for the app to open though.

[patrickjp93]

1 minute ago, Prysin said:

who would use such ancient quad cores?

Many people still hold onto 2500K and 2600K machines, OC of course, but still.

[Prysin]

6 minutes ago, patrickjp93 said:

but still

pathetic.

[patrickjp93]

1 minute ago, Prysin said:

pathetic.

Prove the defense incorrect. The HEDT line isn't called high-end for nothing. They aren't mainstream parts because the mainstream, even in the gaming community, doesn't use them! Now, I can afford to put together a $3000 computer, but I'm also smarter than that considering it's my first year of my first job and my personal goal is to break into the 6-figure salary category. The defense isn't pathetic, especially considering the Steam Hardware survey. My laptop is quite solid, more powerful than many desktops in use today, and has pretty good battery life considering its age, even if it is a MacBook Pro Retina 15".

[DELTAprime]

4 minutes ago, patrickjp93 said:

They aren't mainstream parts because the mainstream, even in the gaming community, doesn't use them!

Ok that's the first thing you've said that I can get behind. Much as I want Ryzen to succeed I won't be buying whatever their 8C 16T part ends up being when it's on the market. Best gaming part if price is no object is the 6700k and soon the 7700k cause cores are not the be all and end all for gaming. I want to know what Ryzen can do against those parts, not the 6900k.

[Prysin]

20 minutes ago, patrickjp93 said:

Prove the defense incorrect. The HEDT line isn't called high-end for nothing. They aren't mainstream parts because the mainstream, even in the gaming community, doesn't use them! Now, I can afford to put together a $3000 computer, but I'm also smarter than that considering it's my first year of my first job and my personal goal is to break into the 6-figure salary category. The defense isn't pathetic, especially considering the Steam Hardware survey. My laptop is quite solid, more powerful than many desktops in use today, and has pretty good battery life considering its age, even if it is a MacBook Pro Retina 15".

read my post, then your post, then my post again...

 

also, 2500k IS a bit pathetic today, even with a decent OC it still only gets close to a FX 8350, because of how games use more threads efficiently then back in the days. 2600k is solid choice, but then again it IS an outrageously outdated i7.

[patrickjp93]

11 minutes ago, Prysin said:

read my post, then your post, then my post again...

 

also, 2500k IS a bit pathetic today, even with a decent OC it still only gets close to a FX 8350, because of how games use more threads efficiently then back in the days. 2600k is solid choice, but then again it IS an outrageously outdated i7.

Other way around. The 8350 gets close to the 2500K.

[Prysin]

14 minutes ago, patrickjp93 said:

Other way around. The 8350 gets close to the 2500K.

Read more reviews Patrick.  Read more reviews 

[patrickjp93]

1 minute ago, Prysin said:

Read more reviews Patrick.  Read more reviews 

I've read all of them.