GOM3RPLY3R :
I have to say, you did a very fine job explaining your self there. And I agree with what your talking about. The only thing that I am going to say is this. After looking over hundreds of benchmarks per GPU and CPU which total add up into the tens of thousands, and after the hundred articles, along with my years of experience, I still see no point in how the 8350 out trims the I7-3770k. 80 to 90% Of the benchmarks I read and after my own testing with the GPUs, I find that it is impossible unless you have very slim variables in which case other hardware allows certain parts to work better than others.
*sigh* The ubuntu benchmarks, which are less biased, show the FX8350
outperforms the i7-3770k in
most categories.
Only matters if you are using ubuntu. Are you using ubuntu? If yes then they are perfectly relevant. If not then throw the benches out.
I am not saying you're wrong, but how do I know that your only really taking the AMD side and posting most things that benefit them? I'm not saying that for certain.
I could ask the same about you...?
The other thing about this is that when I recommend CPUs, I go off of what they're doing, and that is the first question I ask.
But your recommendations are often counter intuitive.
I think both of you are slightly biased. AMD is the better buy at low budgets, no doubt about that especially when you consider it doesn't matter if the inel is getting 130 fps because you are playing on a 60 hz monitor. But If you have the budget, intel is the way to go. The i5 is a very good cpu and in any game that stresses 4 cores to the max it will have a lead over the 8350. But AMD has very nice desktop cpus (mobile is another story) and for a new build should definitely be considered.
After compiling this vast amount of memory and information from my head, it is true and definite that AMD computing is geared more towards video software, multi core apps, and any kind of arcade formatted game like CoD or Starcraft. On the other hand, Intel has a computation geared for heavy work loads, wider range of coded functionality, and benefits greatly when run on more simulated games like ArmA, FSX, and even BF3.
Your examples are backward...AMD runs a tight race in BF3 MP and Crysis 3, FC3, Metro 2033, Bioshock Infinite, etc. Intel wins SC2, Skyrim, and Civ5, CoD (except MP). Also, there is no code for intel that AMD does not incorporate, AMD includes all intel instruction sets;
however,
AMD has instruction sets which are not on intel chips because intel does not integrate AMD instruction sets.
Agree, multicore heavy games run just as well on the 8350 as the i5
Also, many people disregard the true meaning of threaded apps. There's a big dark side to CPUs that of which is like the dark side of the moon. THREADING is the process in which a Processor EXTENDS its total core value. Now these new hypothetical/virtual cores can have more room to work with. This in these simulation games (which are much more threaded than your average game today) BENEFITS from this. The thing that HT will do is either a) create more cores in which that are 'virtual', or b) it will EXTEND each core as if it were much larger. Now, ArmA is a different story as the software for it takes advantage of more cores, BUT the coding for it didn't work when the CPU wants to extend the cores for even better performance than the separate virtual ones, but the game refuses that, and that's where you get the frame decrease. How ever, when HT is shut off (which is not difficult to do) it runs more efficiently on the Intel, even though the AMD still does kick ass with it.
AMD does not have HTT; HTT != real cores. Intel has to divide resources to use HTT, AMD does not have to do that to run multiple threads. So AMD does not "extend core value", whatever you're trying to say. AMD 8 Cores > Intel 4 cores.
If you look at intel slides and demos that is not quite correct. HT essentially uses resources when a process cannot. For example If a program needs to access the RAM (takes years from the cpus perspective), HT allows another process to run on the core instead of the process sitting there waiting for the data. Most of the time HT works, sometimes it does not. In general, HT is powerful (quite efficient from a power and die size perspective) but is geared mainly toward encoding and tasks that do not have a lot of interaction between threads. Games have a lot of interactions and HT performs worse there than one might expect.
Everyone says that AMD will benefit from threaded apps, which cannot be definite as of yet. Same deal with Intel as well. The future will tell if the threaded games will use certain code that benefit one over the other, but right now it lies in the hands of time.
This is what bugs me, you concede above that AMD is better at multithreaded apps, but here you say it's not definite? As though the "thousands of benchmarks" are misleading? AMD + multithreaded apps > intel + multithreaded apps.
As personal opinion based on fact, the games that will really only be threaded are simulation games since they need it. This with what the code is today for it, I can see Intel favoring over it. It also depends on game type. I.e, call of duty works better on AMD because it doesn't require as much code as say ArmA, and also the code and architecture of the game is more simplified and fits into the AMD Kernel better. However with the more complex units of code that come more in large chunks, this is usually where the Intel Kernel exceeds.
AMD kernel? Intel kernel? AMD and Intel are operating systems now? Additionally, I would be willing to bet you the number of lines of code for CoD is more or less the same as the number of lines of code for ArmA. This is another issue I have, you're talking about things you don't know about, and try to make it sound like you do...but you don't. What you said makes
zero sense...
ZERO. I cannot understand any of what you wrote, because it is all self contradicting with references to misused technical terms you clearly don't understand.
You can have a small wheel run X rpm for Y speed, and you can have a big wheel run less than X rpm but the same speed. Think of AMD as the small wheel and Intel as the large. They can both go the same speed but have different rpms. This is where AMD gets their high clock and great resistance from. They need.the extra speed to get extra workload done which in turn makes more heat which needs more heat resistance. Also the fact that a 8 core 8350 can work exactly the same as an I5 and not near as good as the I7 (in most gaming cases) instantly proves that each Intel core is much stronger than the AMD and can sustain them larger workload per core. In a workstation, video, or bit mining perspective, the AMDs cores are stronger for that in opposition. With those kinds of things the code comes basically more stream like rather than a rapid and as Intel is trying to Chuck up this code and get it out, that's where the Intel decreases in performance, while the AMD can easily with the more cores, can stream it out as like it came in. This is what happens with the multiple cores, which basically they are relaxing and minding their own business really.
AMD
modules are stronger than intel
cores, by about 60% if the intel has HTT, if it doesn't they're about 80% stronger. Why do you keep referencing code? Code has
zero to do with CPUs...other than instruction sets being processed. You're trying to tie 2 things together unsuccessfully and cannot talk about either one on a technical level.
This is a bit of an exaggeration. The module structure is at best ~70-75% stronger than one core (and significantly less if if the calculation is FP heavy). Using cinebench (yes bad benchmark but I'm only looking at the scaling not absolute performance, if you can find a better one you can link it) single threaded gets 1.1 points, multithreaded gets 6.9 points. Essentially 6.9-1.1*4= 2.5 2.5/4=63% scaling. HT is around 20% so amd's design is clearly better. Going over the TH review of the 8350 there is NO test where the 8350 beats the i5-3570k by 80% (needless to say, even if there was on average its not that high). I think he means that in code not designed to be parallelized (or which is very difficult to do so for) such as heavily branched code (lots of ifs and elses) which is MUCH harder to parallize the i5 generally wins.
Thus as explained above, if you are doing certain things, pick the cpu that goes with that. Also with tighter budgets, AMDs are the way to go in terms of performance for the more restricted amount of money.
This is what my mind had compiled for the most part. This took about twenty to thirty minutes to type as well, lol. You happy, can I carry on?
I honestly hope you understand how done I am with this conversation...every time you reply, I feel compelled to correct your bad information, and it's nearly driving me nuts trying not to do it. If I read another line talking about AMD's kernel, or Intel's code, my eyes might begin to bleed from the sockets...or my head might explode.
Facebook
Twitter
Instagram