AMD New Horizon

[Energycore]

18 hours ago, That Norwegian Guy said:

6 core or GTFO. That's the optimized count for Star Citizen.

You should play it on a FX-6300 then.

[Dabombinable]

12 minutes ago, Energycore said:

You should play it on a FX-6300 then.

That's only a 3 module 6 thread CPU, and a shit one at that. Its far from a 6 core.

[Energycore]

1 minute ago, Dabombinable said:

That's only a 3 module 6 thread CPU, and a shit one at that. Its far from a 6 core.

Do I have to add /s tags every time?

[AnonymousGuy]

2 minutes ago, Energycore said:

Do I have to add /s tags every time?

For that yeah because so many noobs think AMD "8 cores" are actual 8 cores.

[Dabombinable]

Just now, Energycore said:

Do I have to add /s tags every time?

The majority of people still stay its a 6 core when due to its CMT architecture it isn't, so yes (its one of those things where its technically true, but when I went from a 2.2GHz Phenom II X4 N970 to a "uad core" A8 4555M which boosts to 2.4GHz, fuck AMD because it certainly doesn't perform like it has the stated cores)
 

Spoiler

 

[Energycore]

2 minutes ago, AnonymousGuy said:

For that yeah because so many noobs think AMD "8 cores" are actual 8 cores.

 

1 minute ago, Dabombinable said:

The majority of people still stay its a 6 core when due to its CMT architecture it isn't, so yes (its one of those things where its technically true, but when I went from a 2.2GHz Phenom II X4 N970 to a "uad core" A8 4555M which boosts to 2.4GHz, fuck AMD because it certainly doesn't perform like it has the stated cores)
 

  Reveal hidden contents

 

I'll be that guy and say that if you load the CPU with AVX, it is actually a 6-core

 

But I get all y'all. You can chill.

 

I understand you're salty about that A8-4555M, don't worry the time of CPUs that weak is now done.

[Space Reptile]

@done12many2 heres a thing that might interest you https://docs.google.com/spreadsheets/d/1O2Or6XOETZr3a4gLAuCuYEx8m5X_lv23LdwZiCZpBcM/edit#gid=0

a comparison spreadsheet of the blender times 

[done12many2]

2 minutes ago, Space Reptile said:

@done12many2 heres a thing that might interest you https://docs.google.com/spreadsheets/d/1O2Or6XOETZr3a4gLAuCuYEx8m5X_lv23LdwZiCZpBcM/edit#gid=0

a comparison spreadsheet of the blender times 

 

That is very interesting.  Are all of these using the AMD provided render file used in the comparison demo from New Horizon?  If so, that is an impressive amount of submissions so fast.

 

I also notice they have the Ryzen chip listed as having run the render at 100 samples instead of the default 200.

 

Thanks for sharing that.  I may go ahead and run mine again at 100 now.

[Space Reptile]

3 minutes ago, done12many2 said:

I also notice they have the Ryzen chip listed as having run the render at 100 samples instead of the default 200.

 

yes amd stated that it was run at 100 and not 200 wich is why so many had issues , the sheet is sourced from >here<

[MageTank]

@done12many2 buddy, you are trying way too hard to convince people to think. That's not going to happen on the internet. Sometimes, the most simplistic solution, is the right solution. The simple answer is: Not a single benchmark provided by a company with a conflict of interest; is to be trusted, nor perceived as accurate. If any company; be it AMD, Intel or Nvidia, releases a benchmark claiming X performance, automatically assume it's false until a third party proves otherwise.

 

I understand the desire to test, and try to get early confirmation by comparing your own results, but it's still pointless. Even if you do match or beat it, people will claim that the hardware is not out yet, and still has time to make adjustments. Anyone that has ran blender, should look at those times and automatically question their validity. I don't even own an 8 core SKU, and I question it, lol. 

[Drak3]

53 minutes ago, AnonymousGuy said:

For that yeah because so many noobs think AMD "8 cores" are actual 8 cores.

 

52 minutes ago, Dabombinable said:

The majority of people still stay its a 6 core when due to its CMT architecture it isn't, so yes (its one of those things where its technically true, but when I went from a 2.2GHz Phenom II X4 N970 to a "uad core" A8 4555M which boosts to 2.4GHz, fuck AMD because it certainly doesn't perform like it has the stated cores)
 

  Reveal hidden contents

 

CMT architecture is optimized for workloads that have long threads of instructions. When operated as intended, all 8 cores of a Bulldozer Octocore act as any other Octocore or SMT Quadcore.

AMD's Hexacores are true 6 cored CPUs, their Octocores are true 8 core CPUs. The only people claiming that they aren't, are people that don't understand what actually constitutes a core from a module.

[Drak3]

5 minutes ago, MageTank said:

@done12many2 buddy, you are trying way too hard to convince people to think. That's not going to happen on the internet. Sometimes, the most simplistic solution, is the right solution. The simple answer is: Not a single benchmark provided by a company with a conflict of interest; is to be trusted, nor perceived as accurate. If any company; be it AMD, Intel or Nvidia, releases a benchmark claiming X performance, automatically assume it's false until a third party proves otherwise.

 

I understand the desire to test, and try to get early confirmation by comparing your own results, but it's still pointless. Even if you do match or beat it, people will claim that the hardware is not out yet, and still has time to make adjustments. Anyone that has ran blender, should look at those times and automatically question their validity. I don't even own an 8 core SKU, and I question it, lol. 

I don't question the results, I question how its being reported and how the small details don't always make it out to the public.

 

AMD has nothing to gain from skewing their benchmarks to make ZEN look better. If they did, it'd be a repeat of Bulldozer, except it'd be AMD's fault and not shoddy reporting. It'd effectively push away potential customers, and in all probability, AMD would go under by the end of next year.

 

I have no reason to consider AMD's benchmarks false or skewed, because I have no reason to consider AMD in a place where they could survive the fallout once ZEN released.

[MageTank]

8 minutes ago, Drak3 said:

I don't question the results, I question how its being reported and how the small details don't always make it out to the public.

 

AMD has nothing to gain from skewing their benchmarks to make ZEN look better. If they did, it'd be a repeat of Bulldozer, except it'd be AMD's fault and not shoddy reporting. It'd effectively push away potential customers, and in all probability, AMD would go under by the end of next year.

 

I have no reason to consider AMD's benchmarks false or skewed, because I have no reason to consider AMD in a place where they could survive the fallout once ZEN released.

That would be true, if it hasn't already been done before. With that logic, Intel had nothing to gain from skewing the benchmarks for the Pentium. The simple point is, never take a company for their word. Always wait for reviews from multiple (multiple being the key word here) third party sources. Don't just trust your favorite source, always gather information from multiple sources and compare against each other. 

 

AMD is in a desperate situation. Zen has to not only compete in terms of performance, it has to sell well. AMD cards can compete with Nvidia cards, but they do not sell well, which is a problem. They cannot have this happen for the CPU division. Skewing a result (while not exactly lying, they can always use a disclaimer saying results may vary) is not unheard of when in a desperate situation.

[Space Reptile]

@done12many2

 

 

Edit: just saw that blender says on its own 

 

 

[done12many2]

1 minute ago, MageTank said:

@done12many2 buddy, you are trying way too hard to convince people to think. That's not going to happen on the internet. Sometimes, the most simplistic solution, is the right solution. The simple answer is: Not a single benchmark provided by a company with a conflict of interest; is to be trusted, nor perceived as accurate. If any company; be it AMD, Intel or Nvidia, releases a benchmark claiming X performance, automatically assume it's false until a third party proves otherwise.

 

I understand the desire to test, and try to get early confirmation by comparing your own results, but it's still pointless. Even if you do match or beat it, people will claim that the hardware is not out yet, and still has time to make adjustments. Anyone that has ran blender, should look at those times and automatically question their validity. I don't even own an 8 core SKU, and I question it, lol. 

 

Very well said and very true.

 

I enjoyed the fact that AMD said, something to the effect of "Here, try it yourself and see how fast your rig is compared to our new monster."  I knew right off the bat after my first render that something wasn't right.  

 

You hit the nail on the head.  If it doesn't add up, wait for third parties.

 

I look forward to returning to this thread in a few months after all these Ryzen chips show up.

[done12many2]

19 minutes ago, Drak3 said:

I don't question the results, I question how its being reported and how the small details don't always make it out to the public.

 

AMD has nothing to gain from skewing their benchmarks to make ZEN look better. If they did, it'd be a repeat of Bulldozer, except it'd be AMD's fault and not shoddy reporting. It'd effectively push away potential customers, and in all probability, AMD would go under by the end of next year.

 

I have no reason to consider AMD's benchmarks false or skewed, because I have no reason to consider AMD in a place where they could survive the fallout once ZEN released.

 

Actually, AMD has a great deal to gain.  You can't have a fantastic launch with average performance results.

[done12many2]

8 minutes ago, Space Reptile said:

@done12many2

 

Edit: just saw that blender says on its own 

 

Was that at 100?

[Drak3]

Just now, done12many2 said:

 

Actually, AMD has a great deal to gain.  You can't have a fantastic launch with average performance results.

Bulldozer was skewed by reporters and eMagazines, and now AMD barely scrapes by on the PS4, Xbox One, cheap ass laptops, and GPUs that sell way less than they should.

AMD might gain somewhat in the first month or two, but once they get called out, they can both no longer salvage any reputation they had before Zen, and they lose so many customers that they cannot break even.

 

In the end, AMD has nothing to gain and everything to lose if they skew the results in their favor.

[Space Reptile]

3 minutes ago, done12many2 said:

Was that at 100?

yes

[done12many2]

6 minutes ago, Space Reptile said:

yes

 

Very nice.  2600 is still strutting along!

[Dabombinable]

52 minutes ago, Drak3 said:

 

CMT architecture is optimized for workloads that have long threads of instructions. When operated as intended, all 8 cores of a Bulldozer Octocore act as any other Octocore or SMT Quadcore.

AMD's Hexacores are true 6 cored CPUs, their Octocores are true 8 core CPUs. The only people claiming that they aren't, are people that don't understand what actually constitutes a core from a module.

Only their Phenom II X6 are true hex cores. The rest are far from it as they do not contain all the components of 2x CPU in each module, only some. Modules=1x FPU, 2x ALU. Cores=1x FPU, 1x ALU.

[DocSwag]

55 minutes ago, Drak3 said:

 

CMT architecture is optimized for workloads that have long threads of instructions. When operated as intended, all 8 cores of a Bulldozer Octocore act as any other Octocore or SMT Quadcore.

AMD's Hexacores are true 6 cored CPUs, their Octocores are true 8 core CPUs. The only people claiming that they aren't, are people that don't understand what actually constitutes a core from a module.

Bulldozer octocore =/= SMT quadcore. It's completely different.

[Drak3]

Just now, Dabombinable said:

Only their Phenom II X6 are true hex cores. The rest are far from it as they do not contain all the components of 2x CPU in each module, only some. Modules=1x FPU, 2x ALU. Cores=1x FPU, 1x ALU.

The integer cluster and ALU are the only components that actually make a core. The FPU is a separate, yet vital component.

1 minute ago, DocSwag said:

Bulldozer octocore =/= SMT quadcore. It's completely different.

I said that when operated as intended for a good reason. To imply that they aren't used as intended, and that is why they seem so shitty.

[Dabombinable]

2 minutes ago, Drak3 said:

The integer cluster and ALU are the only components that actually make a core. The FPU is a separate, yet vital component.

I said that when operated as intended for a good reason. To imply that they aren't used as intended, and that is why they seem so shitty.

Pst, most programs make use of the FPU, very few actually use the ALU only now-in fact its been that way since the 386 got replaced by the 486 with its built in co-pro.

[Drak3]

1 minute ago, Dabombinable said:

Pst, most programs make use of the FPU, very few actually use the ALU only now-in fact its been that way since the 386 got replaced by the 486 with its built in co-pro.

That's nice and fine, it doesn't change the fact that AMD made an octocore that had an architecture that went against every method developers used, and it suffered for it.

It doesn't change the fact that the only people claiming that AMD's modules are cores don't know what actually constitutes a core or how CMT is intended to work.