Higher prices, but also bigger dies... what do you think the margins on a 6300 look like?
You don't get it. Until we see EOL on roadmaps for FX83xx and lower , it won't be replaced. There is no logic to do so as these parts command higher price ranges than the APUs.
Why not? Platform is tried and true, very solid. There is zero validation effort for either CPUs or chipsets. They will produce 8T dice for Warsaw product so it's only packaging for AM3 socket that is left(and cost for MCM is much higher than the cost for single socket SKUs). We can only see higher clock or lower TDP with higher yield, a win for both AMD, potential buyers and GF.
There wont be any 8 thead consoles games. Its 6 max for the games.
And then there is the scaling issue. You wont get 50% more performance from 6 cores vs 4.
Higher prices, but also bigger dies... what do you think the margins on a 6300 look like?
Your cost approximation left out R&D and other operating costs, er wait... Hahaha, well played sir.
Your cost approximation left out R&D and other operating costs, er wait... Hahaha, well played sir.
The article in the OP really highlights how ignorant most PC developers are.Matt Higby, creative director of Planetside 2, tells us how game engine optimization on PlayStation 4 will ultimately benefit PC gamers
315mm2 at 300mm Wafer = ~179 dies per wafer
lets say 85% of them can be used = ~152 dies
Your cost approximation left out R&D and other operating costs, er wait... Hahaha, well played sir.
I suspect that 85% figure is a tad optimistic. :hmm:
Wouldn't the performance jump be bigger on intel CPUs? Intel CPUs have better single threaded performance, so if they optimize for 6-8 threads, those of us with 4770k, or the ivybridge/sandybridge equivalents would also see huge boosts right? Probably even bigger than AMD?
I think it'd be interesting to see who the optimizations benefit more. Does it benefit AMD with their cheaper processors that are multithreaded, or intel?
The article doesnt make much sense because no desktops even have 8 core small cat (jaguar) AMD CPUs. Very few AMD CPUs are even 8 cores, and the ones that are have a great deal more power than these low-clocked low-throughput console cores. And to top it off I'm not sure that even if there was an 8 core jaguar chip, it would necessarily run games as fast on a PC OS (windows) due to loss of optimizations inherent to the console OS.
I have a feeling devs will try to limit threads to 4 most of the time when porting to pc regardless. A lot of people game on their laptops after all, and very few of these have more than 4 threads. If the people with 2 core 4 thread intel cpus can't run the games the devs will lose out on a lot of profits.
That's not the way things work. Threads are not allocated 1 per core 100% of the time on any machine built. Just open up windows task manager and you will see how many threads are running on your pc even though it is only a 4 core machine. Right now, for me, there are 1750 threads. Even though MS has said stuff like "1 core for os, 1 for hardware", the actually assignment of threads to physical cores is actually constantly switching back and forth. The os thread never has 100% usage of time. And threads spawned by the game will not all have 100% usage of time either. Therefore, it is necessary to have more threads than the number of available cores in order to actually maximize usage.There wont be any 8 thead consoles games. Its 6 max for the games.
Is it a problem of laziness or have all the easily parallelizable bits been converted to multithreaded code already? Converting the more difficult portions won't give that much more performance.......
I'm a programmer and I know programmers are lazy. Even though multi core has been out for 8+ years now, programmers have not been writing multi threaded code. Even gaming companies have only put minimal multithreading into their engines. The advent of many slow cores will finally force gaming companies to more fully multithread their products.
Is it a problem of laziness or have all the easily parallelizable bits been converted to multithreaded code already? Converting the more difficult portions won't give that much more performance.
Is it a problem of laziness or have all the easily parallelizable bits been converted to multithreaded code already? Converting the more difficult portions won't give that much more performance.
Speaking slightly in the longer term. but probably even medium term, and to an extent now.
We seem to have reached the end of the line, as regards, single core performance, especially as regards to Intels best offerings, but as they catch up (single core speed wise) AMD and Arm.
Barring major technological invention, it seems we can expect very little % increase in single core performance these days, e.g. 5% per year.
And even that 5% is probably diminishing all too quickly.
Therefore in order to move forward, we are going to have to go multicore (as regards SOFTWARE).
So it's not really an Intel vs AMD vs Arm thing, it's more "the laws of Physics", as regards practicable cpu design, of single core performance.
So software is probably going to have to go multicore as much as possible, if it wants to achieve the highest possible computing performance, such as what we want/need for the best/latest games.
So optimization for AMD, is much needed, for ALL cpus anyway, as far as I can see.