Three times the charm - New AMD CPU announcement + big Navi Teaser

[AlwaysFSX]

22 minutes ago, LAwLz said:

Yep, even the 5600X offers better price to performance than Intel's offerings, despite being bad value in my eyes.

At this point AMD is basically only competing with itself, and if I were AMD and my only interest was to make as much money as possible, I'd be kicking myself for releasing a product that was too good last gen (the 3000 series).

I wouldn't be surprised if AMD will stop manufacturing the 3000 series to get it off the market as quickly as possible, so they don't risk cannibalizing sales of the (probably) higher margin 5000 series.

I'm a little bit surprised if they haven't already stopped production of the 3000 series given the 5000 series will be available SoonTM. For the money, they priced these a little bit too high but I assume it's to recoup some R&D money because they know they're getting even more value compared to Intel systems. (Last I looked an Intel Z series board costs more on average than getting something for AMD, savings all around) But they aren't good enough for people on the 3000 series to upgrade at all. Honestly not even for me with my 2600 because I only paid $130 for it.

22 minutes ago, spartaman64 said:

thats not how you should compare performance to price you need to take the cost of the entire system. can you place a 3600 on a table and say to it ok run cinebench? 

 

if you say its not worth to upgrade to the 5000 series if you have a 3000 series cpu then id agree with you but when is it ever worth it to upgrade your cpu in one generation? heck i dont even think its worth it to upgrade your gpu every generation even though gpus have a much higher increase in performance

Incorrect. Total system price is irrelevant because it hasn't changed. Only the CPU price has.

[spartaman64]

4 minutes ago, AlwaysFSX said:

I'm a little bit surprised if they haven't already stopped production of the 3000 series given the 5000 series will be available SoonTM. For the money, they priced these a little bit too high but I assume it's to recoup some R&D money because they know they're getting even more value compared to Intel systems. (Last I looked an Intel Z series board costs more on average than getting something for AMD, savings all around) But they aren't good enough for people on the 3000 series to upgrade at all. Honestly not even for me with my 2600 because I only paid $130 for it.

Incorrect. Total system price is irrelevant because it hasn't changed. Only the CPU price has.

? wym if you buy a more expensive cpu the total system price has increased. comparing cpu prices is pointless since the cpu by itself cant do anything so you always need to spend more to get that performance. its not $299 vs $199 its $999 vs $1099

[leadeater]

11 minutes ago, LAwLz said:

I see a little asterisk there but can't make out what it says. Performance per watt is not a static number and it changes depending on which test and architecture you run.

You can click on it to enlarge it, it's just a see notes slide reference where that information you're asking is contained. If it's anything like the 19% IPC gain it's an average over multiple workloads, I know that is the case for the IPC gain at least.

 

Edit:

Found the slide, note reference you are looking for is R5K-007 so in other words CB R20 nT

 

[LAwLz]

22 minutes ago, spartaman64 said:

thats not how you should compare performance to price you need to take the cost of the entire system. can you place a 3600 on a table and say to it ok run cinebench? 

Of course that's how you should compare performance. That's the only way you get an objective and factual comparison.

Once you start adding the cost of other components then sure the price difference in % goes down, but if you do that then you can't just say it performs 20% better because you won't get 20% higher performance in all tasks.

 

If you're comparing two CPUs, one that costs 200 dollars and one that costs 300 dollars, you can't say the difference in price is only 5% because the rest of your PC costs around 2000 dollars. If you do that, then you can't say you get 20% higher performance because that CPU won't make your GPU better, or your SSD better. 

 

If you're going to look at it from a "whole system cost" perspective then you have to compare for example a 1000 dollar computer with a 5600X vs a 1000 dollar computer with a 3600 and then compare different benchmarks. The 5600X computer will perform better in CPU bound benchmarks, but the 3600 computer could perform better in other benchmarks like maybe GPU bound benchmarks (because it could have a higher end GPU).

 

 

The whole "it's just 5% higher price with a 2000 dollar computer, but it's 20% higher performance" is just a bad way of masking the price difference with other components, but not acknowledging that CPU performance won't matter 100% of the time.

If you say the price isn't 50% higher between the 3600 and the 5600X because you factor in other components, then you are not allowed to say the performance difference is 20% because that does not factor in other components. You can't have it both ways.

 

 

 

29 minutes ago, spartaman64 said:

if you say its not worth to upgrade to the 5000 series if you have a 3000 series cpu then id agree with you but when is it ever worth it to upgrade your cpu in one generation? heck i dont even think its worth it to upgrade your gpu every generation even though gpus have a much higher increase in performance

It's not worth upgrading to the 5000 series regardless of which CPU you got.

My friend has a 2500K. The 5000 series is worse value to him than the 3000 series.

My brother has a Phenom 965. The 5000 series is worse value to him than the 3000 series.

I got a 1700X. The 5000 series is worse value to me than the 3000 series. (Especially to me since I can put a 3000 series in my current motherboard, but the 5000 would require a new motherboard).

 

There is no scenario where the 5000 series is better value than the 3000 series, except if you ONLY care about single threaded performance and do not care at all about multithreaded performance. The same crowd that has kept buying Intel CPUs even though Ryzen was released. This CPU lineup will please those people. Everyone else that has been eyeing or buying Ryzen chips instead of Intel chips? All those people are better off buying the 3000 series over the 5000 series.

[leadeater]

3 minutes ago, LAwLz said:

There is no scenario where the 5000 series is better value than the 3000 series, except if you ONLY care about single threaded performance and do not care at all about multithreaded performance. The same crowd that has kept buying Intel CPUs even though Ryzen was released

That's still a lot of people.

[AlwaysFSX]

6 minutes ago, spartaman64 said:

? wym if you buy a more expensive cpu the total system price has increased. comparing cpu prices is pointless since the cpu by itself cant do anything so you always need to spend more to get that performance. its not $299 vs $199 its $999 vs $1099

Do you really need an explanation that you only compare the two parts that changed?

Do you really need that explained?

[LAwLz]

7 minutes ago, AlwaysFSX said:

Incorrect. Total system price is irrelevant because it hasn't changed. Only the CPU price has.

The way @spartaman64 and some others are calculating it is this way.

 

We have one system with a 3600 that costs 1000 dollars.

If we change the CPU to the 5600X the price goes up to 1100 dollars (because the 5600X is 100 dollars more expensive).

 

Therefore, the $300 5600X is only 10% more expensive than the $200 dollar 3600.

The total price of the PC increases by 10% because the CPU is only a small part of the total system cost.

 

 

The problem (in my opinion) with this is that they then do a 180 turn from "it's about the entire system" and quote a 20% performance increase which is specifically and only about the CPU.

They focus on the entire system regarding the price of the machine, but when it comes to performance they focus on only the CPU.

 

"Oh we're talking about price? Then let's tak about the entire system! We can't just focus on the CPU as an individual component"

"Oh we're talking about performance? Then let's focus on the CPU as an induvial component!"

[AlwaysFSX]

Just now, LAwLz said:

The way @spartaman64 and some others are calculating it is this way.

 

We have one system with a 3600 that costs 1000 dollars.

If we change the CPU to the 5600X the price goes up to 1100 dollars (because the 5600X is 100 dollars more expensive).

 

Therefore, the $300 5600X is only 10% more expensive than the $200 dollar 3600.

The total price of the PC increases by 10% because the CPU is only a small part of the total system cost.

 

 

The problem (in my opinion) with this is that they then do a 180 turn from "it's about the entire system" and quote a 20% performance increase which is specifically and only about the CPU.

They focus on the entire system regarding the price of the machine, but when it comes to performance they focus on only the CPU.

 

"Oh we're talking about price? Then let's tak about the entire system! We can't just focus on the CPU as an individual component"

"Oh we're talking about performance? Then let's focus on the CPU as an induvial component!"

It feels like it's going under the assumption that somehow the price of what you're spending is different even though one component is changed. The motherboard costs the same. So does the monitor. So does the mouse. How do any of those parts calculate the perf/$ of a CPU?

They don't.

 

You're 100% right on this.

[spartaman64]

1 minute ago, LAwLz said:

The way @spartaman64 and some others are calculating it is this way.

 

We have one system with a 3600 that costs 1000 dollars.

If we change the CPU to the 5600X the price goes up to 1100 dollars (because the 5600X is 100 dollars more expensive).

 

Therefore, the $300 5600X is only 10% more expensive than the $200 dollar 3600.

The total price of the PC increases by 10% because the CPU is only a small part of the total system cost.

 

 

The problem (in my opinion) with this is that they then do a 180 turn from "it's about the entire system" and quote a 20% performance increase which is specifically and only about the CPU.

They focus on the entire system regarding the price of the machine, but when it comes to performance they focus on only the CPU.

 

"Oh we're talking about price? Then let's tak about the entire system! We can't just focus on the CPU as an individual component"

"Oh we're talking about performance? Then let's focus on the CPU as an induvial component!"

no like sure cinebench benchmarks are nice to give a general idea but what i really want to see is gaming benchmarks

[spartaman64]

Just now, AlwaysFSX said:

It feels like it's going under the assumption that somehow the price of what you're spending is different even though one component is changed. The motherboard costs the same. So does the monitor. So does the mouse. How do any of those parts calculate the perf/$ of a CPU?

They don't.

 

You're 100% right on this.

because just the cpu itself cant run anything you need the whole system to run something and for the benchmarks i really care about which are gaming benchmarks you benchmark the whole system

[LAwLz]

1 minute ago, leadeater said:

That's still a lot of people.

Not according to the build logs on this forum and Amazon's best sellers list.

I'd consider it a niche.

 

If we look at Amazon for best selling CPUs we see that the first Intel CPU is the 9600K at #8. But people looking at that CPU won't all of a sudden buy the 5600X because they are in completely different price brackets.

If we want to look at how many people are more focused on single threaded performance at all cost, and are willing to spend around 300 dollars or more on a CPU we have to go all the way down to #11 on the list, with the i5-10600K at 275 dollars. If we want to locate the first Intel CPU at over 300 dollars then it's in place #15.

 

So if you ask me, at best this product range will appeal to ~9% of people. The other 91% are better off getting a processor from the 3000 series.

[AlwaysFSX]

2 minutes ago, spartaman64 said:

because just the cpu itself cant run anything you need the whole system to run something and for the benchmarks i really care about which are gaming benchmarks you benchmark the whole system

Annnnnnnnnnnnd you're still wrong.

 

You aren't benchmarking the whole system. At all. You would make the worst person in a scientific field if that's how you think.

You isolate what you're testing so your data isn't flawed. In this case it would be the two different CPUs tested in the exact same system.

Nothing else matters. You do not include the price of anything else.

[spartaman64]

17 minutes ago, AlwaysFSX said:

Annnnnnnnnnnnd you're still wrong.

 

You aren't benchmarking the whole system. At all. You would make the worst person in a scientific field if that's how you think.

You isolated what you're testing so your data isn't flawed. In this case it would be the two different CPUs tested in the exact same system.

Nothing else matters. You do not include the price of anything else.

nice to know how it performs in things i would actually be using it for doesnt matter at all LUL. 

 

also when testing medicine do you put the medicine and bacteria in a Petri dish and say ok it kills the bacteria put it out to market?

no you first do animal tests and eventually run human trials because what you care about isnt whether or not it kills the bacteria but whether or not it cures someone of a disease without killing them. even though those tests introduces more variability and are less "scientific"

[leadeater]

13 minutes ago, LAwLz said:

The problem (in my opinion) with this is that they then do a 180 turn from "it's about the entire system" and quote a 20% performance increase which is specifically and only about the CPU.

They focus on the entire system regarding the price of the machine, but when it comes to performance they focus on only the CPU.

Yes but you also have the same problem in ignoring that there is a performance gain for the system. There will be a point where the 5000 will be giving you more performance than 3000, cases right now and also in the future.

 

You are raising a performance per dollar argument so it's pretty fair game to look at the overall system and how it will perform, yes you could be spending $100 more for the exact same performance per dollar of the system (in general, what ever it really doesn't matter for this point specifically) so it's more a question of do you actually want to spend $100 more or not. You can choose not to, that is a perfectly valid option.

 

There is simply more reasoning right now to just not buy at all, or just limit yourself to a GPU. 5600X is the worst case cost increase and even that difference is not one that gets you to the next GPU tier up, at least not right now anyway.

 

3600 is a really popular gaming CPU so the 5000 series single thread performance gain is of more benefit than more cores on a 3700X or 3900X. Sure ideally there would be a 5600 non x at a lower price but it doesn't exist yet which goes right back to my simple do not buy point. I still however think if you have to and that involves also a new motherboard (not optionally) then 5000 series is the better choice.

[AlwaysFSX]

2 minutes ago, spartaman64 said:

nice to know how it performs in things i would actually be using it for doesnt matter at all LUL. 

You have no idea how benchmarking works. And you always do this where you miss the point of what the data actually means but I'll go ahead and give a short example about why this matters:

You don't mix hardware for CPU performance data because it's dirty data. If this is lost on you then there's no explaining it further. You can put both of these in the same exact system. The only thing that has changed is the price of the cpu ONLY and performance. YOU DO NOT INCLUDE ANYTHING ELSE. IT IS IRRELEVANT DATA THAT SKEWS THE RESULTS.

 

For spending 50% more you need to know the current performance changes.

 

If you want to talk about why it's important that this CPU costs $100 more and talk about system performance:
You're better off buying a 3600 for $100 less and putting that towards a graphics card instead.

 

A 2060 Super MSRP is $400. A 2070 Super is $500.

 

16% better performance stock to stock. You will not see that performance uplift spending $100 more on a CPU at this tier of graphics cards.
Dirty math: i7 8086k@5GHz with 2080 Ti here | Stock 3600 15% below in GN testing with 2080 Ti | 85% 2080 Ti performance ~175fps | 2070 Super at 158.7 is not bottlenecked by the 3600.

[Blademaster91]

28 minutes ago, leadeater said:

Yes but you also have the same problem in ignoring that there is a performance gain for the system. There will be a point where the 5000 will be giving you more performance than 3000, cases right now and also in the future.

 

You are raising a performance per dollar argument so it's pretty fair game to look at the overall system and how it will perform, yes you could be spending $100 more for the exact same performance per dollar of the system (in general, what ever it really doesn't matter for this point specifically) so it's more a question of do you actually want to spend $100 more or not. You can choose not to, that is a perfectly valid option.

 

There is simply more reasoning right now to just not buy at all, or just limit yourself to a GPU. 5600X is the worst case cost increase and even that difference is not one that gets you to the next GPU tier up, at least not right now anyway.

 

3600 is a really popular gaming CPU so the 5000 series single thread performance gain is of more benefit than more cores on a 3700X or 3900X. Sure ideally there would be a 5600 non x at a lower price but it doesn't exist yet which goes right back to my simple do not buy point. I still however think if you have to and that involves also a new motherboard (not optionally) then 5000 series is the better choice.

I agree, I think if you're considering total system cost then Ryzen 5000 would be worth the extra $100 because it's just a difference of say $1000 vs $1,100, if this breaks a budget then do not buy it or wait for prices to lower. A 5800X isn't the best value either, but use a B550 board and it's still cheaper than a 10700K and a decent Z490 motherboard.

[spartaman64]

43 minutes ago, AlwaysFSX said:

You have no idea how benchmarking works. And you always do this where you miss the point of what the data actually means but I'll go ahead and give a short example about why this matters:

You don't mix hardware for CPU performance data because it's dirty data. If this is lost on you then there's no explaining it further. You can put both of these in the same exact system. The only thing that has changed is the price of the cpu ONLY and performance. YOU DO NOT INCLUDE ANYTHING ELSE. IT IS IRRELEVANT DATA THAT SKEWS THE RESULTS.

 

For spending 50% more you need to know the current performance changes.

 

If you want to talk about why it's important that this CPU costs $100 more and talk about system performance:
You're better off buying a 3600 for $100 less and putting that towards a graphics card instead.

 

A 2060 Super MSRP is $400. A 2070 Super is $500.

 

 

16% better performance stock to stock. You will not see that performance uplift spending $100 more on a CPU at this tier of graphics cards.
Dirty math: i7 8086k@5GHz with 2080 Ti here | Stock 3600 15% below in GN testing with 2080 Ti | 85% 2080 Ti performance ~175fps | 2070 Super at 158.7 is not bottlenecked by the 3600.

when did i ever say anything about mixing hardware you used the same gpu motherboard etc etc to test both of the cpus but you need to factor in the cost of the entire system to do any price to performance comparisons since the cpu cant run by itself

 

and like i said when testing medicine do you only test the medicine on the bacteria and nothing else since anything will skew the data? no you test it in humans also since thats its purpose even if it adds more variables you want to test those variables also

[AlwaysFSX]

5 minutes ago, spartaman64 said:

when did i ever say anything about mixing hardware you used the same gpu motherboard etc etc to test both of the cpus but you need to factor in the cost of the entire system to do any price to performance comparisons since the cpu cant run by itself

?????????????????????????????

If it's the same hardware then why would you go by system price. It's irrelevant. That can change. CPU price/performance is what you're testing.

How are variables so difficult for you to understand.
 

[spartaman64]

1 minute ago, AlwaysFSX said:

?????????????????????????????

If it's the same hardware then why would you go by system price. It's irrelevant. That can change. CPU price/performance is what you're testing.

How are variables so difficult for you to understand.
 

because its the system thats running the tests not just the cpu and a 20% increase in cpu cost isnt a 20% increase in system cost

and you are not just buying a cpu you are buying a system

[AlwaysFSX]

Just now, spartaman64 said:

because its the system thats running the tests not just the cpu and a 20% increase in cpu cost isnt a 20% increase in system cost

and you are not just buying a cpu you are buying a system

You are so hung up on "it's the system that runs the tests" instead of:

IT IS THE CPU THAT IS BEING TESTED

 

It can be tested in a $50 board or a $500 board. That is why you cannot add in system price for determining price/performance.

 

I can explain this to you but I cannot understand it for you. You are wrong about testing methodology. Period. There's no discussion about it because every reviewer does not factor in what you think matters.

[spartaman64]

Just now, AlwaysFSX said:

You are so hung up on "it's the system that runs the tests" instead of:

IT IS THE CPU THAT IS BEING TESTED

 

It can be tested in a $50 board or a $500 board. That is why you cannot add in system price for determining price/performance.

 

I can explain this to you but I cannot understand it for you. You are wrong about testing methodology. Period. There's no discussion about it because every reviewer does not factor in what you think matters.

thats why you compare it to the same system i never said anything about changing other parts of the system. tell me how do you run a game with just the cpu you cant so its pointless to talk about it just in terms of the cost of the cpu. 

 

im not even talking about testing methodology im talking about purchasing decisions what part of price to performance is testing methodology that part is already done when you run the benchmark

 

by this logic everyone should just get a 3300x since its the best performance to cpu cost

[AlwaysFSX]

1 minute ago, spartaman64 said:

thats why you compare it to the same system i never said anything about changing other parts of the system. tell me how do you run a game with just the cpu you cant so its pointless to talk about it just in terms of the cost of the cpu. 

 

im not even talking about testing methodology im talking about purchasing decisions what part of price to performance is testing methodology that part is already done when you run the benchmark

 

by this logic everyone should just get a 3300x since its the best performance to cpu cost

You are so hung up on "it's the system that runs the tests" instead of:

IT IS THE CPU THAT IS BEING TESTED

 

It can be tested in a $50 board or a $500 board. That is why you cannot add in system price for determining price/performance.

 

I can explain this to you but I cannot understand it for you. You are wrong about testing methodology. Period. There's no discussion about it because every reviewer does not factor in what you think matters.

 

Purchasing decisions are made by sane testing methodology.

 

I repeat. Again.

 

I can explain this to you but I cannot understand it for you.

[spartaman64]

Just now, AlwaysFSX said:

You are so hung up on "it's the system that runs the tests" instead of:

IT IS THE CPU THAT IS BEING TESTED

 

It can be tested in a $50 board or a $500 board. That is why you cannot add in system price for determining price/performance.

 

I can explain this to you but I cannot understand it for you. You are wrong about testing methodology. Period. There's no discussion about it because every reviewer does not factor in what you think matters.

 

Purchasing decisions are made by sane testing methodology.

 

I repeat. Again.

 

I can explain this to you but I cannot understand it for you.

@leadeater @LAwLz you guys seem to understand what im getting at. im bad at explaining things so please help lol

[The Benjamins]

1 hour ago, spartaman64 said:

@leadeater @LAwLz you guys seem to understand what im getting at. im bad at explaining things so please help lol

you only have to compare system cost in a CPU  test setup if the CPUs motherboard does not have a similar cost version for the other CPU, but typically they do.

 

and if I am builind a system there are some cost I can't avoid like motherboard, ram, PSU, Case they stay nearly static no matter what CPU I am buying. so in the end of the day Price to Performance in games comes down to GPU and CPU cost.

[spartaman64]

7 minutes ago, The Benjamins said:

you only have to compare system cost in a CPU  test setup if the CPUs motherboard does not have a similar cost version for the other CPU, but typically they do.

 

and if I am builind a system there are some cost I can't avoid like motherboard, ram, PSU, Case they stay nearly static no matter what CPU I am buying. so in the end of the day Price to Performance in games comes down to GPU and CPU cost.

thats exactly what allows you to compare the cost of a system to the performance of a cpu if everything else remains the same. 

would you rather buy a 900 dollar computer with a 3300x or a 970 dollar computer with a 3600 given only the cpu changes