AMD VEGA confirmed for 2017 H1

[lordfletch]

Pretty much why I don't bother with this sub forum or thread any more

Doesn't matter what AMD do anymore as it is guaranteed most will always find something to hate about them, if the hardware is fine, then we will see posts about how their drivers suck and nvidias drivers are far better, they get day 1 support for games etc.

Heck, you even have a couple of people who have openly admitted they just want AMD to compete so that they can buy intel/nvidia hardware for cheaper... In some ways, it makes you wish AMD would go down just so that we can see these same people cry about having to pay extortionate amounts for a 5-10% performance increase each year

Good to see that loads is still reeling over his fury(s) purchase too

that burn will never heal (joke)

 

[Rroff]

Integer:

Wonder what handbrake scenario they are using - in all other handbrake tests I've seen my i7 4820K is a clean 30% faster than the AMD 83xx series stock for stock (more like the 4770K v 8370 in the graph for cinebench) and for gaming it just crushes them - unfortunately that integer graph is nothing like an indication of Ryzen's potential gaming performance - in some cases it is struggling against i3s that are basically using 2 real cores and 2 HT units so its going to need a lot more than a scheduler fix as those i3 CPUs are pretty much doing the same thing Ryzen is doing (due to lack of real cores and distribution of resources) in terms of a lot of the heavy work is being done on virtual cores.

 

[levontraut]

Beating nVidia is less important than its price for performance.

As long as the performance is high and the price low it will be a success, combine that with AMD's epic CPU comeback they would provide stiff competition to both Intel and nVidia, incredible when you think that AMD are half the size of nVidia and 4x smaller than Intel.

I have given AMD plenty of tries and I keep getting kicked up the rear with AMD... i Now stay with Intel.. it might be at times a little more expensive but i am staying with Intel.

 

[humbug]

Wonder what handbrake scenario they are using - in all other handbrake tests I've seen my i7 4820K is a clean 30% faster than the AMD 83xx series stock for stock (more like the 4770K v 8370 in the graph for cinebench) and for gaming it just crushes them - unfortunately that integer graph is nothing like an indication of Ryzen's potential gaming performance - in some cases it is struggling against i3s that are basically using 2 real cores and 2 HT units so its going to need a lot more than a scheduler fix as those i3 CPUs are pretty much doing the same thing Ryzen is doing (due to lack of real cores and distribution of resources) in terms of a lot of the heavy work is being done on virtual cores.

Ryzen is not an AMD 38##, whatever that is, Ryzen is a ground up brand new architecture, it shares nothing in common with any of its predecessors, nothing.

You're also misunderstanding the problem with the SMT Scheduler.

Unfortunately the scheduler currently is not able to differentiate principal core threads from virtual SMT threads with Ryzen and in fact sees 16 thread Ryzen 7 processors as processors with 16 physical cores with equal resources per thread.

Because it does not give any preferential prioritization of scheduling tasks to primary threads over SMT threads like it does on Intel platforms, a massively larger percentage of tasks can and do end up getting scheduled for a virtual SMT thread rather than a principal core thread. Resulting in significant artificial performance degradation.

Cinebench is Floating point performance, that is the sort of performance gaming uses, in that the 8c 16t Ryzen hands the 8c 16t Intel its ass. And at less power.

 

[LoadsaMoney]

Another thing, Nvidia don't need to show benchies etc.... of the Ti, as we know its performance already, we've known for nigh on a year (as its just a bigger 1080), as the Ti's are all the same, cut down Titans, less of the vram, but around the same performance, faster, due to having higher clocks, where as Vega, new arc, so absolutely nothing to go on, can't look at the Furys and go, well, Furys were this, Vega has got more of this, this, that, so will be around xxxx better, and also can't go off the 480s.

All we know about Vega is, it can run DOOM at 4k, a smidge faster than the 1080, when its running under Vulkan, but that tells us nothing, as even a potato can run DOOM, and BF1, it can run that well also, being one of their games, it should, but again, tells us nothing, as 290/970, even lower than them cards, can run fine.

With no info, no ones interested, the reduced priced 1070/80s are still going to be flying off shelves, out of warehouses, and the incoming Ti as well, leaving AMD, even less sales, sales of which, they didn't have many of, in the first place, by being over a year late already.

 

[Rroff]

Ryzen is not an AMD 38##, whatever that is, Ryzen is a ground up brand new architecture, it shares nothing in common with any of its predecessors, nothing.

You're also misunderstanding the problem with the SMT Scheduler.

You are misreading my post - how on earth you think someone with my experience would mistake the 83xx series for Ryzen I have no idea. I'm questioning what scenario or validity of those results as from previous handbrake testing I've seen a few of those results don't stack up as I'm used to seeing - which by extension leads me to question how accurately the Ryzen results stack up with the other CPUs.

Regardless of that - as can be seen from other testing handbrake results both single and multi-threaded don't tend to generally reflect performance in anything else and especially aren't a good indication of gaming or potential gaming performance - all cores fully utilised or otherwise.

 

[humbug]

You are misreading my post - how on earth you think someone with my experience would mistake the 83xx series for Ryzen I have no idea. I'm questioning what scenario or validity of those results as from previous handbrake testing I've seen a few of those results don't stack up as I'm used to seeing - which by extension leads me to question how accurately the Ryzen results stack up with the other CPUs.

Regardless of that - as can be seen from other testing handbrake results both single and multi-threaded don't tend to generally reflect performance in anything else and especially aren't a good indication of gaming or potential gaming performance - all cores fully utilised or otherwise.

Roff, every reviewer gets similar Hanbreak results, leave the Tinfoil hat on the role in the kitchen draw.

 

[Rroff]

Roff, every reviewer gets similar Hanbreak results, leave the Tinfoil hat on the role in the kitchen draw.

(EDIT: see second edit below) 3 other sites show handbrake results with a roughly 14% advantage to the 1800X over an i7 7700K in handbrake tests - that one has a 31% advantage - combined with most of the other AMD CPUs showing between 15 and 30% higher results compared to respective Intel CPUs than I normally see - hence wondering how they are testing.

Cinebench might be more like gaming workload in terms of floating point operations, etc. but its different enough that it can easily be synthetically loaded up on every core - hence the almost double performance over the i7 7700K which has half the number of cores of the 1800X - we are a long way if ever from that being reflected in games. The 5960X has relatively low clock rate being designed more for workstation use hence doesn't do so well against the 1800X there - the better clocked 6900K is far closer.

EDIT: My point isn't to diss Ryzen here only to be more realistic as to the gaming performance that people are going to see with Ryzen - it is going to need more than some optimisations in Windows for it to do as well in gaming as it does in other applications.

EDIT2: Looks like there are 2 different schools of testing (maybe due to the BIOS/motherboard issues at launch) for handbrake with Ryzen a few other sites are showing big differences between the 1800X and 7700K (but the other CPUs in their tests are stacking up more like the traditional line up) while some are showing massive differences in Cinebench - even more than Guru3D.

 

[humbug]

3 other sites show handbrake results with a roughly 14% advantage to the 1800X over an i7 7700K in handbrake tests - that one has a 31% advantage - combined with most of the other AMD CPUs showing between 15 and 30% higher results compared to respective Intel CPUs than I normally see - hence wondering how they are testing.

Cinebench might be more like gaming workload in terms of floating point operations, etc. but its different enough that it can easily be synthetically loaded up on every core - hence the almost double performance over the i7 7700K which has half the number of cores of the 1800X - we are a long way if ever from that being reflected in games.

Handbreak has almost an infinite amount of settings and encoding modes, sometimes it doesn't even bother to load up all threads because its not compatible or necessary for the codec. As long as its loading up all threads and your encoding the same file using the same settings you get the same repeatable results.
Despite the higher core count Intel and AMD chips being faster than the 7700K in these tests its not a good idea to compare it to them, the 7700K is a Ghz higher clocked than all of them, its a much smaller high revving engine.

No one expects Ryzen to keep pace with the 7700K in most games, nor do any of Intel's Broadwell chips, people keep forgetting that, its not supposed to, this is the big fat V8 designed to beat Intels big fat V8, and it does that.

Its simply that right now with it being a brand new CPU having no microcode or drivers in Windows its not performing as well with individual cores loaded up as it is with all of those same cores loaded up at once, the Windows patch that is coming will fix that.

PS: the scheduler issue seems to be a Windows 10 thing, probably because it has a lot more Intel optimisations it understands the CPU even less in windows 10 than it does in Windows 7, where this was done...

You look at that game benchmarked in Windows 10 its the exact opposite of that ^^^

 

[Rroff]

No one expects Ryzen to keep pace with the 7700K in most games, nor do any of Intel's Broadwell chips, people keep forgetting that, its not supposed to, this the big fat V8 designed to beat Intel big fat V8, and it does that.

Its simply that right now with it being a brand new CPU having no microcode or drivers in Windows its not performing as well with individual cores loaded up as it is with all of those same cores loaded up at once, the Windows patch that is coming will fix that.

PS: the scheduler issue seems to be a Windows 10 thing, probably because it has a lot more Intel optimisations it understands the CPU even less in windows 10 than it does in Windows 7, where this was done...

Which is the whole point of what I'm saying - don't get hopes up for anything shatteringly better gaming performance even with a fix for how Windows handles it based on how well Ryzen does in some other applications as it simply isn't going to happen.

IIRC Windows 10 has some tweaks with how core parking, etc. works on Kaby Lake which doesn't exist in 7 hence slightly lower gaming performance for those CPUs in 7.

 

[humbug]

Which is the whole point of what I'm saying - don't get hopes up for anything shatteringly better gaming performance even with a fix for how Windows handles it based on how well Ryzen does in some other applications as it simply isn't going to happen.

IIRC Windows 10 has some tweaks with how core parking, etc. works on Kaby Lake which doesn't exist in 7 hence slightly lower gaming performance for those CPUs in 7.

Right, my point was never that it was going to beat Intel's smaller chips in games with the Windows 10 fix, it will not, in the same Broadwell does not, it was simply that Ryzen is under performing compared to how it should be relative to its self with the Scheduler issues resolved.

 

[Rezident]

Hah

Jen's the man for wood screws!! He'll do you a good deal if you ask nice

Heh heh, I'd nearly forgotten about the wood screws. Must find that old YouTube video, that was hilarious.

 

[Kaapstad]

I won't post anything about Vega as I don't want to derail this Ryzen thread.

 

[Rroff]

I won't post anything about Vega as I don't want to derail this Ryzen thread.

Ryzen? - keep up its now a Fermi thread.

 

[N19h7m4r3]

Which is the whole point of what I'm saying - don't get hopes up for anything shatteringly better gaming performance even with a fix for how Windows handles it based on how well Ryzen does in some other applications as it simply isn't going to happen.

IIRC Windows 10 has some tweaks with how core parking, etc. works on Kaby Lake which doesn't exist in 7 hence slightly lower gaming performance for those CPUs in 7.

It really depends on the games. In older games, early 2016 and prior the 7700K wins hands down, even over Intel X99.
Although in more modern games, Ryzen; despite it's scheduling, and motherboard issues is doing rather well.
There is sadly massive variance between reviewers, and most of that is down to motherboards. Gigabyte mobos actually allowing RAM speeds at 3200Mhz, while MSI and Asus struggle on that; and have immature BIOS causing more performance issues.

Golem.de had this to say about their MSI motherboard.

he MSI board was delivered with BIOS version 113, until last Friday a new one appeared.

Version 117, which is still up-to-date, improved speed and stability. If we were still able to count on sporadic Bluescreens with the older UEFI, the board is currently stable. Much more important, however, is the drastically higher performance in games and the real pack with 7-Zip. The release notes include, among other things, a fixed problem with the memory act and its timing as well as the voltage.

Compared to the original bios, the new UEFI increases the image rate in our game course between plus 4 and plus 26 percent, on the average even plus 17 percent!

Benchies in spoilers.

As we can see from these Benchmarks in heavily threaded and modern games Ryzen does really well. We also know it's IPC is just under Broadwell-E.
This means that we should see an increase in gaming performance as Windows 10 is updated, and once the motherboards are finally finished.

I do personally think that AMD should really have waited another month really. Ryzen is fantastic in price to performance, but the entire launch is being marred because of gaming results.

Especially gaming results in older games.

Spoiler

https://www.computerbase.de/2017-03.../#diagramm-battlefield-1-dx11-multiplayer-fps

Battlefield 1

http://pclab.pl/art73043-4.html

 

[humbug]

I won't post anything about Vega as I don't want to derail this Ryzen thread.

Sowy......

 

[SiDeards73]

So, same old same old then? the usual Nvidia detractors bascially hyping up Nvidia products while pointing out AMD have no news, then its derailed by someone and momentum gathers on that with the armchair experts, keyboard warriors and industry specialists all arguing back and forth on stuff they basically know little about other than what they read elswhere on the interwebz!

GPU section at its finest folks

 

[humbug]

Last thing i promise.... just talking about how i agree Ryzen will not beat the 7700K in games even with its Scheduler fix.... along came you with a bunch of computerbase.de slides where in every single one Ryzen is handing the 7700K its ass.... like lol!

It really depends on the games. In older games, early 2016 and prior the 7700K wins hands down, even over Intel X99.
Although in more modern games, Ryzen; despite it's scheduling, and motherboard issues is doing rather well.
There is sadly massive variance between reviewers, and most of that is down to motherboards. Gigabyte mobos actually allowing RAM speeds at 3200Mhz, while MSI and Asus struggle on that; and have immature BIOS causing more performance issues.

Golem.de had this to say about their MSI motherboard.


Benchies in spoilers.

As we can see from these Benchmarks in heavily threaded and modern games Ryzen does really well. We also know it's IPC is just under Broadwell-E.
This means that we should see an increase in gaming performance as Windows 10 is updated, and once the motherboards are finally finished.

I do personally think that AMD should really have waited another month really. Ryzen is fantastic in price to performance, but the entire launch is being marred because of gaming results.

Especially gaming results in older games.

Spoiler

https://www.computerbase.de/2017-03.../#diagramm-battlefield-1-dx11-multiplayer-fps

Battlefield 1

http://pclab.pl/art73043-4.html

 

[N19h7m4r3]

Last thing i promise.... just talking about how i agree Ryzen will not beat the 7700K in games even with its Scheduler fix.... along came you with a bunch of computerbase.de slides where in every single one Ryzen is handing the 7700K its ass.... like lol!

As I mentioned in older games, those released mid 2016 and before the 7700K still pulls a head usually.
Although in nearly all modern games Ryzen is extremely competitive, and the pricing makes them extremely attractive.
Even more so if the Scheduler and motherboard updates at least bring up Ryzen to Haswell-E gaming performance; even though actually IPC is just a smidge under Broadwell-E.

Here are those ones. Total War particularly hates multiple cores over 4. Even the x99 CPUs lose horribly
there.

Spoiler

 

[humbug]

As I mentioned in older games, those released mid 2016 and before the 7700K still pulls a head usually.
Although in nearly all modern games Ryzen is extremely competitive, and the pricing makes them extremely attractive.
Even more so if the Scheduler and motherboard updates at least bring up Ryzen to Haswell-E gaming performance; even though actually IPC is just a smidge under Broadwell-E.

Here are those ones. Total War particularly hates multiple cores over 4. Even the x99 CPUs lose horribly
there.

Spoiler

Yeah, but even if all i did was play pre-2016 games i would still wait on the 6 Core Ryzen, its current performance in old games at worst with a dicky Windows Scheduler is still not at all bad, it will be £100 cheaper than the 7700K and in MT productivity tasks utterly destroy it and with less power consumption.

I you're not excited about what's happening in the world of CPU's right now you're mad, absolutely bonkers. this has been a long time coming.

Oh and Intel, your mainstream chips don't have enough cores....