AMD CPU speculation... and expert conjecture

[juanrga]

Yes, "4C" stands for 4 cores. I used standard formula by AMD performance labs

(CPU Cores x freq x 8 FLOPS)

the 8 FLOPS (per core) follows from 4SP x 2FMA

You can rewrite the formula in term of modules if you want. The result is the same

2M x 16 FLOPS/M x 4.1GHz = 131.2 GFLOP
2M x 16 FLOPS/M x 4.0GHz = 128 GFLOP

You mean

http://accc.riken.jp/secure/4721/07hayashi-amd.pdf
[Appendix A slide 25... at the very end of presentation ]

that is misleading, because it refers to APU v1 or Llano... yet its the same for BD/PD (ver 2 and 3).

Llano had 1 "Fmisc" pipe and 1 "Mul" and 1 "Add" FP pipes... so each pipe was not able of 2 operations (MAD), and what counts is only the "Mul" and "Add" pipes per core, so 4 cores is already 8 FLOP operations, be it Llano or BD/PD( only this last ones have FMAC, more efficient, 2 modules, 4 FMAC pipes, 8 Flop ops in total).

I'm certain is a mistake or "typo" of some kind, 8 flops x 3.5GHz is 28 GFLOPs... and no Llano core i know of... or BD/PD for that matter... can do 8 flop operations per cycle per core in any way( if truth i would like a explanation of how... please) (edt). 8 flops is the number of FP ops per cycle for the "entire APU" with 4 cores.

OTOH, the FLOP number for the GPU is correct: GPU cores x freq x 2 ops (MAD) -> 320 x 920(turbo) x 2 = 588 GFLOPS for Llano... 384 x 875 x 2 = 672 GFLOPs for Trinity (Richland is a little more).

Let me try.

Each 128bit FMAC unit can do 2 DP ops per cycle. The maximum ops are FMA (Fused Multiply Add). Therefore, each unit can do a maximum of 4 FLOPS (2mul + 2add). There are 2 units per Module, two Cores per module

8 FLOPS/M = 4 FLOPS/C

for DP. For SP, it is the double

16 FLOPS/M = 8 FLOPS/C

The same for Intel SB/IB, except that the architecture has double units than AMD. For an i7-3770k

4C x 16 FLOPS/C x 3.5GHz = 224 GFLOP

for SP. For DP, it is one-half. Intel gives only DP figures in their official metrics. i.e. they give 112 GFLOP for its i7-3770k CPU.

TOTAL GFLOP [SP]: CPU + GPU
i7-3770k: 224 + 294 = 518 GFLOP
A10-5800k: 122 + 614 = 736 GFLOP
i7-4770k: 448 + 400 = 848 GFLOP
A10-7800k: 128 + 922 = 1050 GFLOP

 

[juanrga]

I know that in Intel Land (a parallel dimension of Mario World)

now I understand why trolls mention "intel hammer"

i3 3220s are 8 years ahead of FX-8350. The problem is we all live in the real world. In the real world games that made the i3 look good are done. Going forward games are going to be more and more like Crysis 3, and take better advantage of more cores than Crysis 3 did. I think we tend to forget that Crysis 3 is a blueprint of sorts, a first generation game that utilizes many cores. The next generations of those games will utilize the cores more effectively making six and eight core cpus more desirable. If not Intel wouldn't be thinking about releasing an 8 core Haswell (which to the best of my knowledge hasn't happened yet so there is no reason to tout about how great they are until they prove themselves- lord knows 4 core Haswell has had enough troubles 8 core may not release for quite some time).

Everyone knows that Intel has an advantage over AMD right now, but come on a 6 year lead? At least 10 different times hafijur has claimed that Intel is 6 years ahead of AMD in performance. I drink beer and read hafijur's posts and they make no sense, I switch to grain alcohol and just as it seems to become clear... its the next morning the hangover is passing and nothing he's posting makes sense anymore. Seriously I go back and forth wondering if your really naive enough to believe what your posting or if your paid by Intel to post this crap on AMD threads. I've seen a lot of posts from "different" Intel fan boys on different AMD threads and they all have the same exact broken record crap your posting all over here. Either you read all their crap and are re-spreading it or you go by many user names to spread Intel's propaganda.

There is a third option that explains why you see the same crap in each blog/forum open to anonymous posters...

 

[griptwister]

Today, after scrolling through some youtube videos, I came across people saying the Haswell is a hot chip. Then I hear people say the FX 8350 is a hot chip. Then I turned on my Intel Celeron Craptop for old times sake. And you know what, it was HOT!

Conclusion: THEY'RE ALL HOT CHIPS! (Changing the world as we know it with one revolutionary break through)

Haswell chips do run hot, especially when you try to overclock them. The Intel fan club likes to try to dismiss or overlook Haswell's horrible overclocking problems, but they use much more power and run far too hot when overclocked. The 4 core Haswell chips can't handle the vcore load of overclocking (they run far too hot) I have serious doubts about when there will be an 8 core Haswell chip, or what they will have to clock them at to keep the heat under control.

I still think there will be. As long as Intel slightly down clocks the turbo and continues to use the solder for their server CPUs.

 

[Cazalan]

Haswell chips do run hot, especially when you try to overclock them. The Intel fan club likes to try to dismiss or overlook Haswell's horrible overclocking problems, but they use much more power and run far too hot when overclocked. The 4 core Haswell chips can't handle the vcore load of overclocking (they run far too hot) I have serious doubts about when there will be an 8 core Haswell chip, or what they will have to clock them at to keep the heat under control.

No doubt they run hot. Even Ivy runs hot. My i5 K was crashing periodically at stock speeds with the stock HSF. Should have known better to buy an after market HSF on day 1.

I'm sure they will release 8-core versions but they won't be at the 4Ghz speed. That would double the TDP. Look at the sever lines to see the clock speed limits when going to higher core counts.

95W 8-core 2.6Ghz (Ivy)

 

[Mitrovah]

Asus just announced that "later in the year" fm2+ socket motherboards will be for sale. How long does a company usually take between announcing a new motherboard and actually selling a new motherboard? Kaveri is not coming out this year but I have a hopeful hunch the new socket will be available this year nevertheless

 

[noob2222]

something that will make your iq drop by 10%

do yourself and everyone here a favor please.

 

[Guest]

Today, after scrolling through some youtube videos, I came across people saying the Haswell is a hot chip. Then I hear people say the FX 8350 is a hot chip. Then I turned on my Intel Celeron Craptop for old times sake. And you know what, it was HOT!

Conclusion: THEY'RE ALL HOT CHIPS! (Changing the world as we know it with one revolutionary break through)

Haswell chips do run hot, especially when you try to overclock them. The Intel fan club likes to try to dismiss or overlook Haswell's horrible overclocking problems, but they use much more power and run far too hot when overclocked. The 4 core Haswell chips can't handle the vcore load of overclocking (they run far too hot) I have serious doubts about when there will be an 8 core Haswell chip, or what they will have to clock them at to keep the heat under control.

It doesn't make sense to say they use much more power. They literally have a lower TDP & can't overclock V-core as high. The real problem is that the cores can't get the heat out. Sandy Bridge was designed with solder connecting the cores to the CPU cover. For some reason they replaced that direct connection with thermal paste.

from the expert:
http://giatrakis.wix.com/john-giatrakis


I hear that perhaps Intel is bringing back solder with the new enthusiast chip,but that's way out of my price-performance budget. Too bad I sold my Sandy Bridge build before all this came out. Now I'm stuck looking at worse chips than before at a price premium, or waiting and hoping Steamroller has something to offer.

eh,

 

[griptwister]

To be totally honest. Intel is really smart. They're forcing you, (if you're an overclocker) to go with their enthusiast platform. I can guarantee you, those new IB-E x79 CPUs are going to be FAST and OC great! And if not, we'll know 22nm is too small a process. I have a whole theory with the motherboard manufactures, but people would call me crazy. But with the way the z87s are being pushed... it's friggin nuts!

 

[abitoms]

since there is a discussion about heat generation in CPUs, can anyone answer me this question(s) I have in mind for some time?

How much will CPU temperature affect its longevity? I mean will Ivy Bridge and Haswell CPUs die quicker than Sandy Bridges or Piledrivers do? Do reviewers take this into account and test for this longevity?

Haswell might run with lower heat OUTPUT, but as a result with higher temperatures since not a lot of heat is getting dissipated. Is my understanding correct?

 

[griptwister]

since there is a discussion about heat generation in CPUs, can anyone answer me this question(s) I have in mind for some time?

How much will CPU temperature affect its longevity? I mean will Ivy Bridge and Haswell CPUs die quicker than Sandy Bridges or Piledrivers do? Do reviewers take this into account and test for this longevity?

Haswell might run with lower heat OUTPUT, but as a result with higher temperatures since not a lot of heat is getting dissipated. Is my understanding correct?

Lol, CPU temperatures do effect longevity. But by the time your CPU will die, you'll be WAY OVER DUE for an upgrade anyways. The one thing you have to be careful with (this applies to pretty much overclockers only), is Voltage. Too much voltage can kill a CPU. Well, that's what I've heard people say. I personally haven't over-volted my CPU, and I don't intend to. Well, maybe after my upgrade I'll trash my current motherboard and CPU. lol! #inthenameofscience

 

[abitoms]

Lol, CPU temperatures do effect longevity. But by the time your CPU will die, you'll be WAY OVER DUE for an upgrade anyways. The one thing you have to be careful with (this applies to pretty much overclockers only), is Voltage. Too much voltage can kill a CPU. Well, that's what I've heard people say. I personally haven't over-volted my CPU, and I don't intend to. Well, maybe after my upgrade I'll trash my current motherboard and CPU. lol! #inthenameofscience

Thank you for your reply. That was something I did not know.

 

[blackkstar]

Intel currently have laptop cpus turbo to 3.5ghz dual core cpus like the i7 3540m.

If this were address at me I was mentioning that Broadwell 14nm might not be able to hit 3.9ghz turbo. What does 3.5ghz turbo have to do with this?

Intel care about there consumers the most, imagine if intel released slap dash cpus like bulldozer and piledriver.

You mean a new chip that has some regressions, a few really bright spots, but overall was hotter? Yeah, that already happened but instead of pretending it's a desktop chip Intel deflects and says it's all about the mobile, in a sad attempt to avoid the fact that their chips were crap.

If Intel released something like Bulldozer 1 (which I consider Haswell to be very close to), Intel would have reviewers focus on benchmarks that work better on the hardware and they'd push another marketing point besides raw performance.

Oh, and look, the main marketing point for Haswell is power consumption and not performance. Coincidence? Not really.

The reason why intel stopped there ocing (only k series left) chips is simply put the fact of the matter is before you could get lets say a q9300 oc it to 3.7ghz and it easily beats intels fastest quad out.

Congratulations, you don't understand overclocking. The point of being able to do that is that you're skilled enough to actually do it. Do you see the average tablet/facebook class user overclocking their chips? Would you have seen it 5 years ago?

Being able to overclock cheaper chips to beat faster ones was always a kind of a way for Intel and AMD to reward people knowledgeable about their chips. Intel took that away because it was eating their profits. How is that looking out for customers first?

Intel made a logical step allowing turbo boost to high enough levels that without needing to oc you get great performance. Intel went for the multiplier route. A 3% ipc increase is still good, it still is about 100% better then piledriver ipc wise.

You can sit here and talk about how performance is high enough but there are people like me (and my friends) who have 6 core Intels and it still takes hours to accomplish tasks. And now, those of us who would like something faster, have to fight with people like you over the internet because you think all chips should be laptop oriented and focused on mobile at the cost of performance everywhere else in the market.

If it takes me 60 minutes to render something and Intel comes out with a new product, a 3% IPC increase at the same clocks saves me a little under 2 minutes. It's useless to me completely, specially given the fact that I'd basically need at least a $400 investment in a new motherboard + CPU.

8 core haswell is when intel hammer home there significant advantage.

Yeah, I've heard this with GT3 and GT3e, and those chips cost about 5x the price of an AMD APU and about 15x the price of an ARM SoC.

Get ready for 40% faster and 1,500% more expensive!

If amd wasn't so far behind, intel wouldn't have the enthusiast market so one sided, it will get even more one sided when an 8 core haswell cpu which is about 2x faster then the fx9590 while consuming around 60% less power comes out and with new tim heat spreader if it ocs to 5ghz it would be like 3x faster then an fx8350.

8 Core haswell isn't even due for at least a year, we don't even have Ivy Bridge-E yet. FX 9590 won't be around by the time Haswell-E shows up.

Come on man, I don't compare FX 9590 to Q6600 and then say Intel sucks.

If anything amd fanboys should be happy intel leave there performance cpus to there top end expensive line and only have quads for mainstream.

You sure have a knack for posting opinions as facts and then providing no evidence to back them up. No wonder people think you're a troll. Unfortunately, I'm not as nice as them and I view you as mentally incompetent. I'd wager your IQ is around 80 or so and you are somewhat barely functional in society.

It's just fascinating to me that someone with the mental capacity of "HERES MY OPINION I DONT NEED EVIDNECE HERES HOW IT ISSS!" actually got into hardware, usually they're more interested in watching mind-numbing TV shows.

One other thing you think intel improving is bad, amd went backwards with bulldozer how embarrassing as it was hyped to be the next best thing when it was worse then the previous chip on 32nm compared to the 45nm cpu. Intel sandy bridge is more better then probably anything amd will have coming out in next 2 years.

Bulldozer is already a legacy product. Intel has screwed up before (Itanium, Netburst), yet no respectable fanboy ever brings those up because they're in the past and irrelevant to what you can buy today. Even netburst, which blew when it first came out, ended up being kind of good at the end of its life cycle.

Itanium will always blow but it's because Intel blows at nearly everything besides NICs and high performance x86 CPU cores.

 

[jdwii]

blackkstar well said

 

[8350rocks]

something that will make your iq drop by 10%

do yourself and everyone here a favor please.

+1111111111111111111111111111111111111111111111111

 

[8350rocks]

Intel currently have laptop cpus turbo to 3.5ghz dual core cpus like the i7 3540m.

If this were address at me I was mentioning that Broadwell 14nm might not be able to hit 3.9ghz turbo. What does 3.5ghz turbo have to do with this?

Intel care about there consumers the most, imagine if intel released slap dash cpus like bulldozer and piledriver.

You mean a new chip that has some regressions, a few really bright spots, but overall was hotter? Yeah, that already happened but instead of pretending it's a desktop chip Intel deflects and says it's all about the mobile, in a sad attempt to avoid the fact that their chips were crap.

If Intel released something like Bulldozer 1 (which I consider Haswell to be very close to), Intel would have reviewers focus on benchmarks that work better on the hardware and they'd push another marketing point besides raw performance.

Oh, and look, the main marketing point for Haswell is power consumption and not performance. Coincidence? Not really.

The reason why intel stopped there ocing (only k series left) chips is simply put the fact of the matter is before you could get lets say a q9300 oc it to 3.7ghz and it easily beats intels fastest quad out.

Congratulations, you don't understand overclocking. The point of being able to do that is that you're skilled enough to actually do it. Do you see the average tablet/facebook class user overclocking their chips? Would you have seen it 5 years ago?

Being able to overclock cheaper chips to beat faster ones was always a kind of a way for Intel and AMD to reward people knowledgeable about their chips. Intel took that away because it was eating their profits. How is that looking out for customers first?

Intel made a logical step allowing turbo boost to high enough levels that without needing to oc you get great performance. Intel went for the multiplier route. A 3% ipc increase is still good, it still is about 100% better then piledriver ipc wise.

You can sit here and talk about how performance is high enough but there are people like me (and my friends) who have 6 core Intels and it still takes hours to accomplish tasks. And now, those of us who would like something faster, have to fight with people like you over the internet because you think all chips should be laptop oriented and focused on mobile at the cost of performance everywhere else in the market.

If it takes me 60 minutes to render something and Intel comes out with a new product, a 3% IPC increase at the same clocks saves me a little under 2 minutes. It's useless to me completely, specially given the fact that I'd basically need at least a $400 investment in a new motherboard + CPU.

8 core haswell is when intel hammer home there significant advantage.

Yeah, I've heard this with GT3 and GT3e, and those chips cost about 5x the price of an AMD APU and about 15x the price of an ARM SoC.

Get ready for 40% faster and 1,500% more expensive!

If amd wasn't so far behind, intel wouldn't have the enthusiast market so one sided, it will get even more one sided when an 8 core haswell cpu which is about 2x faster then the fx9590 while consuming around 60% less power comes out and with new tim heat spreader if it ocs to 5ghz it would be like 3x faster then an fx8350.

8 Core haswell isn't even due for at least a year, we don't even have Ivy Bridge-E yet. FX 9590 won't be around by the time Haswell-E shows up.

Come on man, I don't compare FX 9590 to Q6600 and then say Intel sucks.

If anything amd fanboys should be happy intel leave there performance cpus to there top end expensive line and only have quads for mainstream.

You sure have a knack for posting opinions as facts and then providing no evidence to back them up. No wonder people think you're a troll. Unfortunately, I'm not as nice as them and I view you as mentally incompetent. I'd wager your IQ is around 80 or so and you are somewhat barely functional in society.

It's just fascinating to me that someone with the mental capacity of "HERES MY OPINION I DONT NEED EVIDNECE HERES HOW IT ISSS!" actually got into hardware, usually they're more interested in watching mind-numbing TV shows.

One other thing you think intel improving is bad, amd went backwards with bulldozer how embarrassing as it was hyped to be the next best thing when it was worse then the previous chip on 32nm compared to the 45nm cpu. Intel sandy bridge is more better then probably anything amd will have coming out in next 2 years.

Bulldozer is already a legacy product. Intel has screwed up before (Itanium, Netburst), yet no respectable fanboy ever brings those up because they're in the past and irrelevant to what you can buy today. Even netburst, which blew when it first came out, ended up being kind of good at the end of its life cycle.

Itanium will always blow but it's because Intel blows at nearly everything besides NICs and high performance x86 CPU cores.

+1

 

[-Fran-]

Lol, CPU temperatures do effect longevity. But by the time your CPU will die, you'll be WAY OVER DUE for an upgrade anyways. The one thing you have to be careful with (this applies to pretty much overclockers only), is Voltage. Too much voltage can kill a CPU. Well, that's what I've heard people say. I personally haven't over-volted my CPU, and I don't intend to. Well, maybe after my upgrade I'll trash my current motherboard and CPU. lol! #inthenameofscience

The first part is correct, but the second part not quite.

Voltage doesn't kill any kind of circuit. It's the Amperage who kills stuff because of resistance that turns into heat melting stuff or changing it's conductive properties. If you're within the thermal properties of the metal/substance inside the CPU, then you can have your CPU running for an eternity with no issues. So, in short, OC'ing is not dangerous because of Voltage itself, but because of the power running around (VA -> wattage) turning into heat changing chemical properties of the circuitry of the CPU (transistors could melt, stop conducting or even reverse conductivity in some cases). I know that you will argue "but Amps are constant when we use more Volts". That's true, but that translates to making more joules go around the circuitry upping the potential difference for the circuit. It's like the "a gun doesn't kill a person, a person kills a person" type of argument, haha.

Cheers!

 

[daerohn]

To both intel and AMD fans. It's crap to say AMD is way behind intel, intel rules, AMD sucks. stop this nonsense. What AMD did was to change the entire foundation of the CPU with BD. In fact this is why they bought ATI in the first place. AMD is trying to integrate GPU and CPU in a single die. Why? Have you ever tried to use yıour GPU for calculations like video encoding? I have and I would say that the performance increase is dramatically high. Intel on the other hand remained on the old CPU architecture making it more power efective and faster.

AMD cpu's are not bad, yes they can suck in IPC and power effieceny yet this is a new technology which has just born. When compared to standard x86 architecture which has been here for tens of years now it is a baby. This technology needs time to evolve. When it is mature it will be far more ahead of intel. In fact it is already ahead of Intel in multi core performance.

For intel supporters, you can have any intel chip you can and boast about their performance, however the very basic of intel cpu's are derived from the patent licences they purchases from AMD in the past. And I remember when intel was the only CPU manufacturer. The prices back then was too high and competition with AMD forced them to make better chips with lower prices. For me Intel chips are still too expensive for what they offer. There is another issue with Intel. They cheat. Yes, and they cheat in a very bad manner. All the tests, compilers out there favors Intel. They also paid great deal of money to OEM's not to produce any AMD based computers in the past. The real question you should ask is "why does Intel need to cheat?". In fact to gain more money, this means emptying your pockets (so why you still support a company who are trying to rob you?). AMD on the other hand is strugling with tight budgets and doesn't have the resources Intel has to force OEM's to support them more. However they made a good move with Sony and Microsoft to use their APU's for next gen consoles. With this neither MS nor game developers ignore them anymore. We will see more softwares, games in the future that are optimised for AMD. And guess what in the next few years, with compilers optimized for AMD the tides will turn we will see the truth behind Intel's lies.

The very first reason I usually buy AMD is I do not like cheaters, and I hate to see the big bastards making fun of weak ones. Intel cheats, tell lies and this must end.

 

[griptwister]

Lol, CPU temperatures do effect longevity. But by the time your CPU will die, you'll be WAY OVER DUE for an upgrade anyways. The one thing you have to be careful with (this applies to pretty much overclockers only), is Voltage. Too much voltage can kill a CPU. Well, that's what I've heard people say. I personally haven't over-volted my CPU, and I don't intend to. Well, maybe after my upgrade I'll trash my current motherboard and CPU. lol! #inthenameofscience

The first part is correct, but the second part not quite.

Voltage doesn't kill any kind of circuit. It's the Amperage who kills stuff because of resistance that turns into heat melting stuff or changing it's conductive properties. If you're within the thermal properties of the metal/substance inside the CPU, then you can have your CPU running for an eternity with no issues. So, in short, OC'ing is not dangerous because of Voltage itself, but because of the power running around (VA -> wattage) turning into heat changing chemical properties of the circuitry of the CPU (transistors could melt, stop conducting or even reverse conductivity in some cases). I know that you will argue "but Amps are constant when we use more Volts". That's true, but that translates to making more joules go around the circuitry upping the potential difference for the circuit. It's like the "a gun doesn't kill a person, a person kills a person" type of argument, haha.

Cheers!

Thank you! This is what I'm here for, correction and learning! Other people would be glad to know this too. I'm in favor of second amendment rights! Lol.

 

[noob2222]

Lol, CPU temperatures do effect longevity. But by the time your CPU will die, you'll be WAY OVER DUE for an upgrade anyways. The one thing you have to be careful with (this applies to pretty much overclockers only), is Voltage. Too much voltage can kill a CPU. Well, that's what I've heard people say. I personally haven't over-volted my CPU, and I don't intend to. Well, maybe after my upgrade I'll trash my current motherboard and CPU. lol! #inthenameofscience

The first part is correct, but the second part not quite.

Voltage doesn't kill any kind of circuit. It's the Amperage who kills stuff because of resistance that turns into heat melting stuff or changing it's conductive properties. If you're within the thermal properties of the metal/substance inside the CPU, then you can have your CPU running for an eternity with no issues. So, in short, OC'ing is not dangerous because of Voltage itself, but because of the power running around (VA -> wattage) turning into heat changing chemical properties of the circuitry of the CPU (transistors could melt, stop conducting or even reverse conductivity in some cases). I know that you will argue "but Amps are constant when we use more Volts". That's true, but that translates to making more joules go around the circuitry upping the potential difference for the circuit. It's like the "a gun doesn't kill a person, a person kills a person" type of argument, haha.

Cheers!

On that note the current of a cpu definately isn't a constant by any means. Transistors are constantly turning on and off and at any given time. Its when you have too many "on" that can damage a cpu. This is what degrades a cpu over time, having one lead transistor in a serices circuit being constantly overworked. Generally a single transistor can take X current over a certain period of time, keeping it "on" causes it to heat up. At the designed specs, the transistors won't "self destruct".

A transistor is "on" when its told to "do something" you are correct it isn't the voltage being applied, but rather the "overclock" keeping the "on" state longer in that same period of time. Voltage only allows the transistors to properly see the "on" fast enough.

 

[GOM3RPLY3R]

The hardware performance depends on IPC, clock speed, and number of cores. Increasing any of those you increase performance in general.

When did this happen ? Has AMD developed something new that automagically converts single threaded code to multithreaded ? Maybe they could make it public and open source it, unlike the greedy Intel bastards.

BTW, you haven't done much real world coding, have you ?

Honestly, there is one thing where you can manually change the cores it uses: ArmA.

In the cfg. file, you can set how many cores to recognize. It's scale starts at one, but is easily corresponded from lowest to highest (0-1, 1-2, 2-3 and so on). Now if Hyper Threading did help the game and I had a 3930k, then I would have it register 12 cores, and then it utilizes it. However, the windows settings override anything, so if Windows doesn't allow it, then the game can't receive it.

 

[GOM3RPLY3R]

You know what, honestly, this thread should just shut down. There's really no point in going on when only ~ 10 pages of the 123 are actual steamroller information. Why not just make an article that updates and people can subscribe to them, or even better yet, just let people look for it their own way! I've been trying to get off this thread and help people on real threads, but i have to keep coming back to it, since there's always something going on. I mean seriously! Just post a Steamroller thread when it does come out, so people can talk about all it's specs, what needs to be done better, etc. This shit seriously needs to stop!

 

[mayankleoboy1]

The first part is correct, but the second part not quite.

Voltage doesn't kill any kind of circuit. It's the Amperage who kills stuff because of resistance that turns into heat melting stuff or changing it's conductive properties. If you're within the thermal properties of the metal/substance inside the CPU, then you can have your CPU running for an eternity with no issues. So, in short, OC'ing is not dangerous because of Voltage itself, but because of the power running around (VA -> wattage) turning into heat changing chemical properties of the circuitry of the CPU (transistors could melt, stop conducting or even reverse conductivity in some cases). I know that you will argue "but Amps are constant when we use more Volts". That's true, but that translates to making more joules go around the circuitry upping the potential difference for the circuit. It's like the "a gun doesn't kill a person, a person kills a person" type of argument, haha.

Cheers!

AFAIk, excess voltage destroys the CPU itself by degrading the electron-hole concentration of the semiconductor itself.

 

[GOM3RPLY3R]

The first part is correct, but the second part not quite.

Voltage doesn't kill any kind of circuit. It's the Amperage who kills stuff because of resistance that turns into heat melting stuff or changing it's conductive properties. If you're within the thermal properties of the metal/substance inside the CPU, then you can have your CPU running for an eternity with no issues. So, in short, OC'ing is not dangerous because of Voltage itself, but because of the power running around (VA -> wattage) turning into heat changing chemical properties of the circuitry of the CPU (transistors could melt, stop conducting or even reverse conductivity in some cases). I know that you will argue "but Amps are constant when we use more Volts". That's true, but that translates to making more joules go around the circuitry upping the potential difference for the circuit. It's like the "a gun doesn't kill a person, a person kills a person" type of argument, haha.

Cheers!

AFAIk, excess voltage destroys the CPU itself by degrading the electron-hole concentration of the semiconductor itself.

Both of you are kind of not on the right page. I'll make it simple. Would it make sense to run a stock FX-8350 @ 2.0 V-core? No, that fucker would be fried before it got to the loading screen for Windows. For example: basically, that's why for the 3930k, I'm aiming at ~ 1.25 V-core for 4.0 Ghz, and ~ 1.28 for 4.5 Ghz, when the stock is 3.2 Ghz @ 1.20 V-core. Many people are saying to run it @ 1.4 V-core minimum for 4.0 Ghz, but really, it'd be dead in less than a month if I did that. I read an article where someone was pushing their 3770k to the limits, and they just pulled off 1.16 V-core @ 4.5 Ghz, instead of the minimum recommended 1.2 V-core. So he safely set it at 1.17, and had no problems ever since. It even runs ~2-4 ºC cooler at that V-core than 1.2. ^_^

 

[thaares]

Should i buy fx-8320 now, or wait for steamroller. Do you guys thing Vishera with microsoft hotfix will be very close to preformance of steamroller?

 

[juanrga]

Should i buy fx-8320 now, or wait for steamroller. Do you guys thing Vishera with microsoft hotfix will be very close to preformance of steamroller?

The hotfix offers from about a 5% increase in half-threaded (~ 4 threads) applications to 0% in single and full-threaded applications (>= 8 threads). Not a chance.

I suppose that you mean "wait for Steamroller FX". My opinion is that there is no such chips:

i)
AMD server roadmap show that the 12--16 core piledriver Opteron chips are being substituted by Warsaw chips with 12--16 'tweaked' piledriver cores, whereas the 4--8 core piledriver Opterons chips are being substituted by 4 steamroller core Berlin. Warsaw chips could be substituted by 6 core steamroller chips or maybe by 4 core excavator chips (nobody knows).

FX series has been traditionally derived from server series. Therefore I don't wait a steamroller FX 8500 series.

ii)
AMD said we would wait one more FX generation after the 8300 series and they released the 9000 series. Therefore I don't wait a FX 8500 series.

iii)
AMD is trying hard on HSA APUs. Note that the Kaveri/Berlin APUs with HSA enabled software would be about 3x faster than a FX-9000 series CPU for certain workloads. A steamroller FX 8500 series would be a big step backward.

Others in this thread think otherwise and wait a steamroller 8500 series...