Nerevarine
Incarnate
The AMD 8xxx series is supposed to be a rebadge of the current 7xxx series
Should i upgrade my GPU or wait for next gen cards?
they already did that for laptop oem but not for desktop. They'll introduce completely different chip for desktop 8000 series. I checked in ocn forums, there is a thread on it with all the specs mentioned. Check it out.
Hrishi
******************
A decent suggestion would be to Slightly Overclock your card using a beginners guide to Overclocking GPUs/CPUs. It's not that hard really , as it sounds unless you are planning on something huge like 30-40% overclock or more.
If you having enough cash to spare , then that's a different story.
The setup you have isn't that old.It'll suffice for an year. Upgrading right now wouldn't be as clever as waiting for an year and then upgrade. New generation of games are coming , so there's a good probablity that current GPU cards will get obsolete in a li'l while. So , wait for those games to be released and then choose your card accordingly.
I went through an article , which stated that next-gen games are going to be graphic intense.
If you having enough cash to spare , then that's a different story.
The setup you have isn't that old.It'll suffice for an year. Upgrading right now wouldn't be as clever as waiting for an year and then upgrade. New generation of games are coming , so there's a good probablity that current GPU cards will get obsolete in a li'l while. So , wait for those games to be released and then choose your card accordingly.
I went through an article , which stated that next-gen games are going to be graphic intense.
overclocking wont help much if not done aggressively It will result in only 5-10% performance increase. Not enough to compensate for what a high end gpu can offer. Next gen games are more likely to use the console power to achieve higher fps at higher resolutions without any drastic change in the look of the games. That being said current gen of gpu already run games on these levels and will do in future titles too. Next gen gpu wont provide a huge performance boost either it will be around 10-20% but lower power consumption which has become a major challenge for gpu makers these days as they are more likely to compete in mobile sector.
So all in all you are safe in the hands of current gen gpus for atleast 2-3 years.
So all in all you are safe in the hands of current gen gpus for atleast 2-3 years.
lol .. most of the good games should have different versions for pc and console(s) ( instead of some stupid buggy porting stuff ) and pc version always looks better as it requires a pc with high specs which can kick out any console in terms of processing power be it a cpu or gpu and no console is ever gonna be more powerful than pc.
Cilus
laborare est orare
Topgear, now the concept is little different. Currently in PC, CPU and GPU can't work over a single item in a uniform manner, it is either dedicated to CPU or GPU. But in the new consoles, AMD implemented a new type of Heterogeneous Computing, namely HSA or Heterogeneous System Architecture which unifies all the Processing elements like CPU, GPU and any other co-processors under a single block and Programmers or developers can write their program in such a manner with the API provided that the application can be executed by all the available processing resourcse, not by only CPU or GPU or special purpose elements. It is also providing HUMA or Heterogeneous Unified Memory Access which enables both CPU and GPU to access same memory space, Page Tables and Virtual Memory space smae way, resulting reduction of latency and ease in programming.
Because of those facts, the current generation consoles can deliver performance in such a level shich might be higher for only CPU or only GPU used in current generation PC hardware. Non to mention, consoles run very optimized and efficient OS which is tailor made for their hardware, resulting very efficient management and hence higher performance.
I have just started learning OpenCL programming language which is considered as the future language for Heterogeneous System and belive me, now getting ideas about how HSA will work. The ideas are although similar to a x86 based high level language, the implementations are quiet different.
Because of those facts, the current generation consoles can deliver performance in such a level shich might be higher for only CPU or only GPU used in current generation PC hardware. Non to mention, consoles run very optimized and efficient OS which is tailor made for their hardware, resulting very efficient management and hence higher performance.
I have just started learning OpenCL programming language which is considered as the future language for Heterogeneous System and belive me, now getting ideas about how HSA will work. The ideas are although similar to a x86 based high level language, the implementations are quiet different.
nikhilsharma007nikx
Wassup?
WTF !! man !! u truly are a geek !!!!
avinandan012
Cyborg Agent
If only someone invents something which can bypass DirectX in Windows then we can have very efficient games written for PC with high visual quality still enjoyable in a Mid range PC without HSA.
Waiting for a day in future where a Linux based alternative will be available for gaming PCs.
Waiting for a day in future where a Linux based alternative will be available for gaming PCs.
well he really is.
Hrishi
******************
Its advised to get high on VRAM as much possible. GDDR5.Hmmmmm so i guess buying a new card next year would be better
And a CPU with as many cores.
i5 2500K with some high card like 7970 won't be enough?
i mean i can upgrade GPU but CPU too seems kinda costly
Hrishi
******************
Yes , it should suffice. What I am talking about is higher amount of Graphic Memory , like say 4 GB GDDR5.i5 2500K with some high card like 7970 won't be enough?
Thanks Cilus for being a great source of knowledge always
even I had high hopes from linux ( used to play windows games on linux too ) but still pc gaming is DX based and will remain like this on the future.
dan4u
Took Off!!!
You mean like AMD's APU series...
harshilsharma63
DIY FTW!
Yes, its like APU.
Cilus
laborare est orare
Although current generation APUs ( Llano, Trinity and Richland) provides very good scope for Heterogeneous computing, they are not HSA and don't support Unified Memory Access or UMA. The advantage here is that CPU cores and GPU dies ae placed insde a single package, resulting very fast switching among them and assigning tasks more efficiently as per the need. But they memory architecture they use is called Unified Memory Architecture which is different from Unified Memory Access.
Let me explain a little:-
In the current generation APU, the GPU die accesses the same Physical Memory which the CPU is also using. But here, for GPU, an amount of Ram is allocated as a logical block which CPU can't access and can only accessed by GPU. Similarly the rest of the Memory can only be accessed by CPU and GPU can't access it. Say I have 8 GB Ram in my system and 2 GB is allocated for GPU. So inside an APU, GPU can only access that 2 GB portion and CPU can only access the rest 6 GB portion. Now suppose a program consisting of three threads, say T1, T2 and T3 are placed inside the memory accessed by CPU.
T1 & T2: CPU thread
T3: Graphics Thread, needs GPU.
Now how things will work in current generation APU:
1. CPU will fetch information about T1, T2 and T3.
2. 1st it will assign T1 and T2 to its cores.
3. For T3, it will make a request to GPU die by message passing techniques and share the information of T3 (Like the data where it will be operated, instructions).
4. Based on the information, GPU will allocate a Space in within its 2 GB memory. If required size is not available, GPU might increase the allocated Ram size dynamically or remove some data which is not currently in use. It might take several CPU clock Cycle.
5. After that, CPU will copy the tasks and the data required for T3's execution into Memory allocated to GPU. It will also maintain a lock to its own copy of T3's data to that other programs can't modify it while T3 is being processed in GPU.
6. Then GPU will start executing T3.
7. After execution is completed, GPU will save the updated data or result into its own allocated memory where T3 was copied.
8. It will then sends a signal to CPU, informing the completion of the task and the location where the result is stored.
9. CPU will then copy from that data from memory allocated to GPU to its own memory and updates its own copy of T3.
Now if HUMA is implemented, there won't be any logical distinction CPU memory or GPU memory and they will access the same memory in a unified manner. So how a upcoming Kaveri APU will execute the same task:
1. CPU will fetch information about T1, T2 and T3.
2. 1st it will assign T1 and T2 to its cores.
3. For T3, it will pass the information, starting address and end address of T3 to GPU.
4. GPU will directly fetch T3 from Ram and start executing it. It also maintain a lock to it so that other CPU programs can't access it.
5. Once the thread T3 is completed, GPU will directly update the result in Ram and again transfer the control to CPU.
I think now you guys can understand in a high level how the program will be faster even without increasing the performance of CPU or GPU alone. In PS4 and XboX Next SOC, AMD has already implemented HUMA (Heterogeneous Unified Memory Access) and Kaveri will be the 1st PC APU to implement it. Here even a modest performance boost in either CPU or GPU or both can result very high performance boost in overall capabilities due to these advanced features. OpenCL is being considered the the next generation language for programming and HSA is being implemented in every aspect, even in Mobile SOC platform. Qualcom, Samsung, Imagination (PowerVR GPU makers), MediaTek, Texus Instruments, ARM, LG...all are members of HSA foundation and working to improve performance in their SOC (Which is just like an APU, having multiple different type of Processing units like CPU, GPU, Digital Signall Processor, Wifi Module etc integrated in a single chip). Recently Qualcom has shown OpenCL based Image editing applications and Gaming performance in Kindle Fire HD where the improvement was more than 50% in some cases.
So you understand that how HSA model is going to be the future...very powerful CPU or GPU will not hold the future...it will be unified performance all the way.
OpenCL mod for the Kindle Fire HD reveals untapped graphics potential (hands-on video)
Let me explain a little:-
In the current generation APU, the GPU die accesses the same Physical Memory which the CPU is also using. But here, for GPU, an amount of Ram is allocated as a logical block which CPU can't access and can only accessed by GPU. Similarly the rest of the Memory can only be accessed by CPU and GPU can't access it. Say I have 8 GB Ram in my system and 2 GB is allocated for GPU. So inside an APU, GPU can only access that 2 GB portion and CPU can only access the rest 6 GB portion. Now suppose a program consisting of three threads, say T1, T2 and T3 are placed inside the memory accessed by CPU.
T1 & T2: CPU thread
T3: Graphics Thread, needs GPU.
Now how things will work in current generation APU:
1. CPU will fetch information about T1, T2 and T3.
2. 1st it will assign T1 and T2 to its cores.
3. For T3, it will make a request to GPU die by message passing techniques and share the information of T3 (Like the data where it will be operated, instructions).
4. Based on the information, GPU will allocate a Space in within its 2 GB memory. If required size is not available, GPU might increase the allocated Ram size dynamically or remove some data which is not currently in use. It might take several CPU clock Cycle.
5. After that, CPU will copy the tasks and the data required for T3's execution into Memory allocated to GPU. It will also maintain a lock to its own copy of T3's data to that other programs can't modify it while T3 is being processed in GPU.
6. Then GPU will start executing T3.
7. After execution is completed, GPU will save the updated data or result into its own allocated memory where T3 was copied.
8. It will then sends a signal to CPU, informing the completion of the task and the location where the result is stored.
9. CPU will then copy from that data from memory allocated to GPU to its own memory and updates its own copy of T3.
Now if HUMA is implemented, there won't be any logical distinction CPU memory or GPU memory and they will access the same memory in a unified manner. So how a upcoming Kaveri APU will execute the same task:
1. CPU will fetch information about T1, T2 and T3.
2. 1st it will assign T1 and T2 to its cores.
3. For T3, it will pass the information, starting address and end address of T3 to GPU.
4. GPU will directly fetch T3 from Ram and start executing it. It also maintain a lock to it so that other CPU programs can't access it.
5. Once the thread T3 is completed, GPU will directly update the result in Ram and again transfer the control to CPU.
I think now you guys can understand in a high level how the program will be faster even without increasing the performance of CPU or GPU alone. In PS4 and XboX Next SOC, AMD has already implemented HUMA (Heterogeneous Unified Memory Access) and Kaveri will be the 1st PC APU to implement it. Here even a modest performance boost in either CPU or GPU or both can result very high performance boost in overall capabilities due to these advanced features. OpenCL is being considered the the next generation language for programming and HSA is being implemented in every aspect, even in Mobile SOC platform. Qualcom, Samsung, Imagination (PowerVR GPU makers), MediaTek, Texus Instruments, ARM, LG...all are members of HSA foundation and working to improve performance in their SOC (Which is just like an APU, having multiple different type of Processing units like CPU, GPU, Digital Signall Processor, Wifi Module etc integrated in a single chip). Recently Qualcom has shown OpenCL based Image editing applications and Gaming performance in Kindle Fire HD where the improvement was more than 50% in some cases.
So you understand that how HSA model is going to be the future...very powerful CPU or GPU will not hold the future...it will be unified performance all the way.
OpenCL mod for the Kindle Fire HD reveals untapped graphics potential (hands-on video)
Jaskanwar Singh
Aspiring Novelist
steam for linux?
Featured Linux Games
quite a few good games are available on linux now, more are coming in future.