Don'T quote me on that, but afaik it is like this:
The 8350 has 4 so-called piledriver modules. Each module consists of 2 integer cores that share 1 floating point unit and the L2 cache. The L3 cache is shared among all modules.
In comparison, Intel HT works with 1 Integer core that has 2 floating point units and the L2 cache is shared among all cores.
Why do something simple, when there is a complicated way?
Ryzen 7 2700X | BeQuiet Dark Rock Pro 4 | 16GB DDR4-3200 | MSI X470 Gaming Pro | MSI GTX 1070 Gaming X 8G | 500GB / 750GB Crucial SSD
Fractal Define C | LG 32UK550 | Das Model S Professional Silent | CM Storm Xornet
What makes these guys less reliable than anandtech? Or tom's hardware? Or any other reviewing site? Pretty much nothing. TTL would've been worthwhile because when he runs his own benchmarks he often just leaves the camera running while it's going, and showing the results of it, but his only video on the processor was speaking around the subject, so we don't have any more validation method for him either.
On the subject on the motherboard, it really shouldn't matter as long as the CPU is stable. If people know there's an issue with the specific motherboard it may very well be worth looking into, but otherwise I'd just shut up if the processor overclock is stable.
Last edited by Drunkenvalley; 2013-01-26 at 07:25 PM. Reason: Derp'd. Had some shoddy writing.
Dumbed-down version in plain english goes something like:
Intel's Hyperthreading means there's another ~5% core in addition to every full core. Tiny programs can run on this fraction of a core with very highly optimized math problems completely self-contained. Otherwise normal threads such as games will only use the HT cores for faster task switching which happens hundreds of times every second.
AMD's Bulldozer modules start as two complete cores for the integer part, but after that the floating point and multimedia sections are shared between the two cores. Some old-timers might remember original Intel processors in 80's and early 90's before Pentiums came in two parts. The main processor and math co-processor. Basically what AMD has done is that there's two main processors, one for each core, but they share the math co-processor. If the program fits completely in the integer part (like you see in compression benchmarks) Bulldozer modules show their best performance.
So in theory AMDs Bulldozer is better system, but the problem why it's not doing well in benchmarks is really simple. Vast majority of programs rely heavily on floating point math. Tasks like video compression is 95% running on the shared math part and it can't be moved to the integer core which explains why multitasked programs aren't really getting full advantage of the processor. Another half of the problem is AMD's lower performance per core which means single threaded programs are slower.
edit:
What AMD is trying to do is the APUs where the main processor runs integer code, and the GPU part runs floating point math, essentially going back to Intel's 80's model with two separate processors on same chip. Why this does not work in reality is that there isn't enough software support yet to run the math part of plain vanilla code in the GPU but instead it will be executed in the math-gimped CPU.
Last edited by vesseblah; 2013-01-26 at 08:04 PM.
Never going to log into this garbage forum again as long as calling obvious troll obvious troll is the easiest way to get banned.
Trolling should be.
"They OC the AMD chip over 500mhz higher than the intel one and then proceed to say that a 4.5ghz clock for the i5 is excellent, amazing etc."
Sorry but the simplest fact is this guy is as stupid as he claims the guys teksyndicate are. Cpu frequency is not linear it is not identical from cpu to cpu, so saying that a 5.0Ghz AMD chip has some advantage over a 4.5Ghz i5 because of its 500Mhz frequency is just down right wrong. Frequency is not necessarily a speedometer for CPU's, as old Core2's run at 3.1-4Ghz and have NO where near the performance of lower clocked current CPUs.
4.5Ghz does not hold a speed value like 45 MPH. Still imo an i5-3570k will be the better choice overall for a casual gamer. This cannot be said about college students in graphical arts, computer sciences, and or hardcore gamers who record, stream , and or edit video. The 8350 would be the optimal dollar choice, as an i7-3xxx would be better by a percentage here and there but the $100~ is up to you and the wallet you have.
Current AMD line is great for people utilizing many cores and high processor speeds (Rendering, After Effects, 3DS Studio Maxa, Maya ETC).
However Intel CPU's across the board have higher results in games, simply because of the better use of low cores (1-2).
That's a bit dodgy to say, Daverid. Presumably that would be what you saw in games, yes, but frankly an issue is that in many cases, the processor doesn't even matter all that much in the first place.
While it's true that many games don't use many cores, many of them don't even really take advantage of the hardware anyway. In this scenario, you could argue that the additional processing power that can be scheduled to run simultaneously would be more beneficial, which shows when you for example stream in full HD or other such silly things.
Streaming isn't really all that popular though... And I was just simply stating "Better Results in Games in General". There's definitely some games which are outliers, and perhaps something like Streaming+Gaming would be better with an AMD. But since Streaming is a very rare thing to do, I wouldn't really think it's worth mentioning when it comes to just 'General Gaming'. And really that's why the 3570k is always recommended, because it's better for just General Gaming and Computer Use.
Not really sure why people even feel they need 8 cores for this...
http://www.twitch.tv/deltruss/b/318008836
Fast forward to 26~ minutes in, but the folks who were actively watching the stream said it was superb quality.
I'm not saying IPC doesn't matter, but it's definitely something to consider when dealing with a variety of titles. It's also probably noteworthy that newer titles are seeing more multithread performance as well, far as I've seen and heard, which hopefully also closes the gap as well.
In no way am I going to advocate not giving a shit or anything about the game's CPU performance, but it's something to bear in mind when people wonder why AMD are more on the "let's add more cores" train of thought. Being able to schedule more simultaneous instructions should definitely matter when dealing with smaller, but continuously running tasks that actively require CPU time.
Intel's got the sheer IPC to mostly just deal with it by chucking it onto the cores' stacks, so it's largely a moot point though, I guess.
And i pointed out what's wrong about it. Remind me to not spend 2hrs doing benchmarks again if noone reads them...