AMD on the road to recovery.

[4K8KW10]

So AMD up the core count massively in the mainstream and you hate on them because their entry level mainstream £175 CPU 'only' has 6 cores, with SMT!

Have you looked at the crap Intel are still flogging in the sub £200 bracket lately? https://www.overclockers.co.uk/pc-components/processors/intel/core-i5

look at this 4/4 junk for £150 https://www.overclockers.co.uk/inte...ocket-lga1151-processor-retail-cp-64s-in.html

intel has bad offers, yes. They run the risk of being thrown out of the market or lagging deeply behind because of this so called "out-of-date" technology. They stimulate and keep their sales by influencing the market and trying to convince (where using normals means, where with shady ones) us that their offers are good enough.

My idea is that AMD should not wait them and try to run ahead as far as possible, so making the probability that intel might catch sometime in the future highly unlikely.
Mr. Papermaster said in the interview that AMD's strategy is independent of what intel offers now.

"We set out a roadmap that would bring AMD back to high performance and keep us there. It is independent of our competitors roadmaps and semiconductor node execution on 10nm. And we'll continue to drive our roadmap in that way. We called a play, we've been executing as we called it, and that's what you're going to see at AMD, just tremendous focus on execution. If we do that, then it is less about focusing on our competition, and about being the very best we can be with every single generation."
https://www.tomshardware.com/news/a...ores-coming-in-the-era-of-a-slowed-moores-law

 

[humbug]

intel has bad offers, yes. They run the risk of being thrown out of the market or lagging deeply behind because of this so called "out-of-date" technology. They stimulate and keep their sales by influencing the market and trying to convince (where using normals means, where with shady ones) us that their offers are good enough.

My idea is that AMD should not wait them and try to run ahead as far as possible, so making the probability that intel might catch sometime in the future highly unlikely.
Mr. Papermaster said in the interview that AMD's strategy is independent of what intel offers now.

"We set out a roadmap that would bring AMD back to high performance and keep us there. It is independent of our competitors roadmaps and semiconductor node execution on 10nm. And we'll continue to drive our roadmap in that way. We called a play, we've been executing as we called it, and that's what you're going to see at AMD, just tremendous focus on execution. If we do that, then it is less about focusing on our competition, and about being the very best we can be with every single generation."
https://www.tomshardware.com/news/a...ores-coming-in-the-era-of-a-slowed-moores-law

The reality is Intel couldn't care less about what you or i think about the products you and i buy, the bulk of their business comes from data centre and OEM, they are still ahead of AMD in mobile because for some inexplicable reason AMD aren't executing Zen 2 + Navi already.

In the Data centre Intel are lagging behind AMD, look at TR3 vs Icelake, that's basically the situation in the data centre, much higher core dencity, better per core performance, better power consumption, right now that's where Intel are putting everything they have into stopping AMD running away with what is Intel's core and most important business because if they aint out there begging, manipulating, cheating that sector into staying with Intel thier revenues fall off a cliff and AMD find themselves suddenly swimming in what was Intel's money mountains.

 

[Chrisc]

So AMD up the core count massively in the mainstream and you hate on them because their entry level mainstream £175 CPU 'only' has 6 cores, with SMT!

Have you looked at the crap Intel are still flogging in the sub £200 bracket lately? https://www.overclockers.co.uk/pc-components/processors/intel/core-i5

look at this 4/4 junk for £150 https://www.overclockers.co.uk/intel-core-i3-8300-3.7ghz-coffee-lake-socket-lga1151-processor-retail-cp-64s-in.html

What £150 use is that ^^^ to anyone? That's a £99 CPU.

Its interesting now that we're in a position where a 6c/12t CPU may only last 5yrs tops.(im not saying it will, its purely hypothetical) if games start to become heavily dependent on higher core/thread count. Do we run the risk of going back to the late 90s early 2000's when hardware changed at a rapid rate?

 

[henson0115]

Its interesting now that we're in a position where a 6c/12t CPU may only last 5yrs tops.(im not saying it will, its purely hypothetical) if games start to become heavily dependent on higher core/thread count. Do we run the risk of going back to the late 90s early 2000's when hardware changed at a rapid rate?

not really since games still have settings that can be adjusted up and down the cpu stack.

 

[4K8KW10]

The reality is Intel couldn't care less about what you or i think about the products you and i buy, the bulk of their business comes from data centre and OEM, they are still ahead of AMD in mobile because for some inexplicable reason AMD aren't executing Zen 2 + Navi already.

In the Data centre Intel are lagging behind AMD, look at TR3 vs Icelake, that's basically the situation in the data centre, much higher core dencity, better per core performance, better power consumption, right now that's where Intel are putting everything they have into stopping AMD running away with what is Intel's core and most important business because if they aint out there begging, manipulating, cheating that sector into staying with Intel thier revenues fall off a cliff and AMD find themselves suddenly swimming in what was Intel's money mountains.

Are there any advertising campaigns done by AMD in which they explain the benefits of having more cores in OEM noteboks, for example?
If they tell the users that their Google Chrome needs 8 cores to function normally and not lag, then there would be a higher chance to open more work and market presence for them there.

I mean, there should be stronger marketing efforts, not just Cinebench and framerates in games.

 

[humbug]

Its interesting now that we're in a position where a 6c/12t CPU may only last 5yrs tops.(im not saying it will, its purely hypothetical) if games start to become heavily dependent on higher core/thread count. Do we run the risk of going back to the late 90s early 2000's when hardware changed at a rapid rate?

What i was surprised to see is just how quickly games started using high levels of Multithreading pretty soon after AMD dropped the 8 core mainstream on us, new consoles will get 8 core 16 thread Zen 2, no, i'm not expecting my 3600 to be good on a higher end GPU for more than another two years, i'm ok with that, the good thing about sticking to around £200 ish CPU's i can get a brand new one every two years, i cycle GPU > CPU + stuff > GPU > CPU + stuff... but this time the 1070 is proving a tough nut to crack, its over due an upgrade but i'm just not tempted by anything just yet.

 

[humbug]

Are there any advertising campaigns done by AMD in which they explain the benefits of having more cores in OEM noteboks, for example?
If they tell the users that their Google Chrome needs 8 cores to function normally and not lag, then there would be a higher chance to open more work and market presence for them there.

I mean, there should be stronger marketing efforts, not just Cinebench and framerates in games.

Unfortunately AMD and marketing don't mix well, they are far better at doing than talking, with Intel it clearly looks like its the other way round.

 

[Perfect_Chaos]

Does anyone else think AMD was kind of misleading in their graphs showing overall performance in games? When you see videos from Digital Foundry they say themselves that AMD can be as much as 10-30% behind (worst case) in gaming, at least that shows in some of their tests anyway. But I guess you can't say it's misleading if AMD were showing average FPS charts, but we all know they don't tell the full story.

Hoping the 4000 chips give a good overall boost. Bit confused why other sites that have done tests don't seem to show as much of a difference as Digital Foundry have, not that I don't trust them though.

 

[humbug]

Does anyone else think AMD was kind of misleading in their graphs showing overall performance in games? When you see videos from Digital Foundry they say themselves that AMD can be as much as 10-30% behind (worst case) in gaming, at least that shows in some of their tests anyway. But I guess you can't say it's misleading if AMD were showing average FPS charts, but we all know they don't tell the full story.

Hoping the 4000 chips give a good overall boost. Bit confused why other sites that have done tests don't seem to show as much of a difference as Digital Foundry have, not that I don't trust them though.

I don't know about 30 to 40%, that to me seems extreme, i have never seen anyone make claims Zen 1 being 30 to 40% behind, that's like Bulldozer performance and i don't trust it, its a massive outliar compared to the rest of the internet. That seems deliberately at odds with the rest for click bait.

Lets take 6 of the main reviewers

Jayz2cents, Hardware Unboxed and TPU are in the overall 95% of Coffeelake performance range, GN, Linus and Toms are in the 90% range, GN with some of his are at about 85%

I think those 6 are all right. It also depends on how you test them, Hardware Unboxed and JayZ use better quality RAM than GN and Toms.

Then theres the games, if you're only testing a few games and they are all like Farcry 5 / Newdawn is to Ryzen, like what Steve from GN does, then you would think its about 85% of coffeelake, if like Hardware Unboxed you have a much broader pallet of game, like 37 including FC5 then overall it is around 95% as Ryzen 3 really only lags behind in about 4 or so games out of 10 times that many.

You can make a case one way or the other in what you as a reviewer do with it, i feel like Steve prefers to air on the side of "still has a some catching up to do" where as Steve from Hardware Unboxed is more like "heres just about every sort of game you're likley to use it for and this is the result"

Digital Foundry - <40%? no, i don't buy that.

 

[humbug]

I don't know about 30 to 40%, that to me seems extreme, i have never seen anyone make claims Zen 1 being 30 to 40% behind, that's like Bulldozer performance and i don't trust it, its a massive outliar compared to the rest of the internet. That seems deliberately at odds with the rest for click bait.

Lets take 6 of the main reviewers

Jayz2cents, Hardware Unboxed and TPU are in the overall 95% of Coffeelake performance range, GN, Linus and Toms are in the 90% range, GN with some of his are at about 85%

I think those 6 are all right. It also depends on how you test them, Hardware Unboxed and JayZ use better quality RAM than GN and Toms.

Then theres the games, if you're only testing a few games and they are all like Farcry 5 / Newdawn is to Ryzen, like what Steve from GN does, then you would think its about 85% of coffeelake, if like Hardware Unboxed you have a much broader pallet of game, like 37 including FC5 then overall it is around 95% as Ryzen 3 really only lags behind in about 4 or so games out of 10 times that many.

You can make a case one way or the other in what you as a reviewer do with it, i feel like Steve prefers to air on the side of "still has a some catching up to do" where as Steve from Hardware Unboxed is more like "heres just about every sort of game you're likley to use it for and this is the result"

Digital Foundry - <40%? no, i don't buy that.

Actually you know what let me put it this way.

I tested my own 3600 against Toms Hardwares result in FC5, its a canned benchmark, not much you can do to it....

So Toms said 82 FPS for the 3600, at stock with XMP 3000 CL16 i got 86, meh... margin of error but still that looks a little low to me.

So i set my Ram to 3333Mhz and tuned the timings a little to make it more like it would be with Ryzen 3000 rated 3200Mhz B-Die

I ended up with 94 FPS, and that with the GPU topping out at the end.

So you see how easy it is to say one thing, or another about its performance.

Use crap RAM with lose timings... use the box cooler and bingo! a true measure of its performance, but also not necessarily representative.

 

[Perfect_Chaos]

https://youtu.be/gGARjN5M158 go to 19:30 he mentions 10-30% a bit after that

 

[4K8KW10]

Actually you know what let me put it this way.

I tested my own 3600 against Toms Hardwares result in FC5, its a canned benchmark, not much you can do to it....

So Toms said 82 FPS for the 3600, at stock with XMP 3000 CL16 i got 86, meh... margin of error but still that looks a little low to me.

So i set my Ram to 3333Mhz and tuned the timings a little to make it more like it would be with Ryzen 3000 rated 3200Mhz B-Die

I ended up with 94 FPS, and that with the GPU topping out at the end.

So you see how easy it is to say one thing, or another about its performance.

Use crap RAM with lose timings... use the box cooler and bingo! a true measure of its performance, but also not necessarily representative.

Your resolution is 1280 x 720.
The difference between 82 and 86 is big enough to cause changing of places with respect to a competing product.

 

[Boot_Loop]

We should purchase CPUs with more cores, preferably 8, 12 and more, while in return AMD will benefit with more sales of its new technology, while we will get better user experience - higher productivity levels, more life-like gaming environments, modern not out-of-date technology, etc.

I can certainly get behind that. Imagine if progs were optimized properly to use 8-16 cores and 16-32 threads. Worker productivity would shoot up massively.

 

[humbug]

Your resolution is 1280 x 720.
The difference between 82 and 86 is big enough to cause changing of places with respect to a competing product.

Yes to remove the GPU bottleneck.

 

[Chrisc]

not really since games still have settings that can be adjusted up and down the cpu stack.

I'd hope so

 

[Chrisc]

What i was surprised to see is just how quickly games started using high levels of Multithreading pretty soon after AMD dropped the 8 core mainstream on us, new consoles will get 8 core 16 thread Zen 2, no, i'm not expecting my 3600 to be good on a higher end GPU for more than another two years, i'm ok with that, the good thing about sticking to around £200 ish CPU's i can get a brand new one every two years, i cycle GPU > CPU + stuff > GPU > CPU + stuff... but this time the 1070 is proving a tough nut to crack, its over due an upgrade but i'm just not tempted by anything just yet.

I'm exactly the same with the Vega, I want a tad more oomph as I game at 1440p but the 5700xt doesn't quite give me enough of a jump.

 

[henson0115]

I'd hope so

most games devs wouldn't want to loose the sales from the lower end of the market, considering this is most likely the biggest or closest to biggest part of the market. so there will always be a minimum specification which will be fairly low down the cpu stack. the 90's are long gone where developers made games without any kind of analytics. i wouldnt worry about this at all to be honest. yes the low end goal posts might move but they will still be catered for in games.

 

[Chrisc]

most games devs wouldn't want to loose the sales from the lower end of the market, considering this is most likely the biggest or closest to biggest part of the market. so there will always be a minimum specification which will be fairly low down the cpu stack. the 90's are long gone where developers made games without any kind of analytics. i wouldnt worry about this at all to be honest. yes the low end goal posts might move but they will still be catered for in games.

Oh I'm not worried, merely generating a hypothetical question. Personally I think we will be alright but with the consistent push for more cores its interesting to hear peoples thoughts

 

[4K8KW10]

Yes to remove the GPU bottleneck.

I was under the impression that you would like to showcase discrepancies between the results on your own system and the ones in the review...

most games devs wouldn't want to loose the sales from the lower end of the market, considering this is most likely the biggest or closest to biggest part of the market. so there will always be a minimum specification which will be fairly low down the cpu stack. the 90's are long gone where developers made games without any kind of analytics. i wouldnt worry about this at all to be honest. yes the low end goal posts might move but they will still be catered for in games.

Well, it again depends on the marketing and convincing the users to upgrade to something modern.

Oh I'm not worried, merely generating a hypothetical question. Personally I think we will be alright but with the consistent push for more cores its interesting to hear peoples thoughts

We had been on quad core CPUs from the launch of Core 2 Quad Q6600 on January 8, 2007 till the launch of the Ryzen 7 1800X on March 2, 2017.
What we merely got was just some hyper-threading, or virtual doubling of the cores to eight virtual processors.
That was 10 years of stagnation. It's like we lost these 10 years for nothing.

The highlight of this decade is Crysis 3, and even today, 7 years later after its launch, there are still no games that dramatically improve the visuals, if any modern games improve the visuals at all, to begin with.

 

[4K8KW10]

AMD: Our Goal is to Get 10% Server Share by Q2 2020 with 2nd Gen EPYC CPUs & to Surpass Historical Levels in Both Server & Desktop Segments

"AMD acknowledges that while their current server market share is about 7%, it's their goal over time to get back to the historical market share levels of 26% that they once had with their older Opteron server processors. But before they reach that, they have to set smaller yet still significant targets. Their current goal is to get to a double-digit (10%) server market share by Q2 2020 and we are already a few weeks away from Q1 2020."

https://wccftech.com/amd-goal-historical-cpu-market-share-server-desktop-notebook-segments/