you realize you just basically said that multi card configurations ought to avoid an AMD set up?
- Status
you realize you just basically said that multi card configurations ought to avoid an AMD set up?
"Then again, software fixes for hardware problems are only viable when software was the problem originally. I remember once telling a programmer that his computer had a bad memory module. Rather than swapping it out, he charged in with determination to create a software-based solution. Had he identified the bad memory cells and kept his system from accessing them, he might have enjoyed about as much success as AMD waiting for Windows 8.
I eventually talked the programmer into fixing the hardware problem, rather than doggedly looking for a never-quite-finished software solution. AMD, do you see where I’m going with this?"
Lol i did exactly this on one of my computers. Except i never replaced the ram, it still has the bad stick in it(there are 3 bad 4k pages on the stick), and the software solution i wrote, works just fine to keep it from crashing....
I eventually talked the programmer into fixing the hardware problem, rather than doggedly looking for a never-quite-finished software solution. AMD, do you see where I’m going with this?"
Lol i did exactly this on one of my computers. Except i never replaced the ram, it still has the bad stick in it(there are 3 bad 4k pages on the stick), and the software solution i wrote, works just fine to keep it from crashing....
There are examples of software fixes for hardware flaws all over the place. Look at the Linux bootup kernel messages, it will mention which ones are being applied. Broken BIOSs, broken networking chips, CPU bugs, USB chip flaws, etc. Don't even get started on video drivers and what silicon issues they have to compensate for. Just like software, almost all hardware has flaws in it.
I repeatedly see posts that bash AMD which are built on a foundation of ignorance. You people need to get educated before you go criticizing CPU design you know absolutely nothing about. Simply latching onto a benchmark number and using it as a basis for stating that a product "sucks" is missing the forest for the trees.
I repeatedly see posts that bash AMD which are built on a foundation of ignorance. You people need to get educated before you go criticizing CPU design you know absolutely nothing about. Simply latching onto a benchmark number and using it as a basis for stating that a product "sucks" is missing the forest for the trees.
DJDeCiBeL
Splendid
- Dec 3, 2010
- 6,060
- 1
- 27,965
I don't dispute that at all, but when you're talking about the piss poor IPC of BD, software has no chance of fixing that. That's purely a hardware based restriction. No bit of "optimized" code will reverse that, no matter how much AMD fanboys may hope and pray that it will.
Thankfully, Piledriver is better (not enough, though), and Steamroller will be be even better, still (should finally be truly competitive in single threaded tasks with current Intel CPU's).
That's beside the point of this article, though.
Sorry, but all this talk about "software not being able to help hardware" issues seems so close-minded to me. Time and time again, software can be adapted to take advantage of different, new hardware systems.
GPGPU for example, also multi-threading. The fact that we see patches that improve performance, driver updates. Come on people! Just because AMD and/or MS didn't make something good of this up to now doesn't mean it isn't absolutely impossible. I have heard of ways to mod BD (with software) to adapt to software that aren't that well-threaded. And I have heard about AMD's slip-ups with their implementation of BD and how PD fixes some of them, though that doesn't mean enthusiasts can't try to clean-up after these slip-ups. Whether or not it's "too much of a hassle" or "not worth it," as long as it is possible, it shouldn't be said to not exist!
I'm not really an AMD fanboy here (I use an Intel system), but I am a "fanboy" against close-mindedness...
GPGPU for example, also multi-threading. The fact that we see patches that improve performance, driver updates. Come on people! Just because AMD and/or MS didn't make something good of this up to now doesn't mean it isn't absolutely impossible. I have heard of ways to mod BD (with software) to adapt to software that aren't that well-threaded. And I have heard about AMD's slip-ups with their implementation of BD and how PD fixes some of them, though that doesn't mean enthusiasts can't try to clean-up after these slip-ups. Whether or not it's "too much of a hassle" or "not worth it," as long as it is possible, it shouldn't be said to not exist!
I'm not really an AMD fanboy here (I use an Intel system), but I am a "fanboy" against close-mindedness...
esrever
Splendid
- Dec 27, 2011
- 5,386
- 0
- 26,160
[citation][nom]BigMack70[/nom]Can this put an end to the nonsense from AMD fans that Bulldozer is this amazing future CPU and that our opinions of it are going to change in the future?No they're not. Bulldozer was a fail; Piledriver is at least a step in the right direction away from fail, but the ridiculous claims of AMD's CPU fans need to be put to rest. The problem is AMD, not Windows or anything else, and AMD needs to step up their game on the CPU front.[/citation]Although bulldozer isn't an amazing architecture, there is improvement in threading and such happening on all fronts of software, just looking at the Bulldozer review on toms vs the Piledriver review shows that with very small changes that ended up being a 10-15% improvement showed the architecture in a completely new light. The majority of the improvement from piledriver was from hardware but some of it was also in software, in some programs like winrar, the latest versions are much better threaded than the one a year ago.
There is inherent flaws in every cpu design. Bulldozer had quite a few of them but its unfair to say that the hardware is completely at fault since hardware is always made to run software which isn't static. One of the fundamental thing about cpu design is to know that software evolves just as hardware, no benchmark is timeless and always accurate. Although people still run superpi as a benchmark, it no longer correlates to any real world measurements. This happens with everything, there is a reason people don't do benchmarks of today's cpus on windows 95 and it will be plainly apparent if you check out results of software from that era why.
Heres in regard to your other posts. Why would you not want a 5000 word essay with charts? whos going to read 1 sentence about something and give it credit, this way we get data and a point of view and in my opinion is a great thing. Having articles like this is much more interesting than reading a one liner saying "NO". Its not about convincing fanboys but more about being professional and writing articles which are interesting and backed with credible data. your rant to wisecracker didn't make sense to me either but Im not getting into that since I didn't really follow how his comment related to the article.
BTW SLI/Crossfire rigs work perfectly fine on AMD, it has the same flaws as a single gpu setup and nothing more. AMD isn't recommended for gaming with a single gpu if you want the best fps either but the games are still fully playable, same would be said about crossfire/SLI rigs.
There is inherent flaws in every cpu design. Bulldozer had quite a few of them but its unfair to say that the hardware is completely at fault since hardware is always made to run software which isn't static. One of the fundamental thing about cpu design is to know that software evolves just as hardware, no benchmark is timeless and always accurate. Although people still run superpi as a benchmark, it no longer correlates to any real world measurements. This happens with everything, there is a reason people don't do benchmarks of today's cpus on windows 95 and it will be plainly apparent if you check out results of software from that era why.
Heres in regard to your other posts. Why would you not want a 5000 word essay with charts? whos going to read 1 sentence about something and give it credit, this way we get data and a point of view and in my opinion is a great thing. Having articles like this is much more interesting than reading a one liner saying "NO". Its not about convincing fanboys but more about being professional and writing articles which are interesting and backed with credible data. your rant to wisecracker didn't make sense to me either but Im not getting into that since I didn't really follow how his comment related to the article.
BTW SLI/Crossfire rigs work perfectly fine on AMD, it has the same flaws as a single gpu setup and nothing more. AMD isn't recommended for gaming with a single gpu if you want the best fps either but the games are still fully playable, same would be said about crossfire/SLI rigs.
Well if you go by reviews, a 5~10% difference in performance is the difference between utter garbage and pure gold. It was the difference between the HD3870 and the 8800GT. Everyone bought the GT because of a 5~10% difference. So a 0~2% difference always helps.
how many more attempts will it take to admit reality?
people can talk about patches until they are blue in the face but what matters is results. and so far the outcome has fallen terribly short of the speculated, desired or hoped to achieve results.
lets have some fun here. you have $100 that you need to invest. someone tells you they can give you a 15% increase if you let them invest it for you. you give them the money but later they come back with $105 (5%) but not $115. you go ok, i can deal with that. then they tell you you can get another 15% if you invest again with them and you try but this time they a get return of not $120.75(15%), not $110.25 (5%) but $106.05 (1%)
so instead of $132.25 (15%+15%)you are holding 106.05 (5%+1%). you gonna have faith in that investor anymore? (i would be looking for another investor ASAP)
(open can of worms)
lets say that windows 8 does get a "magic bullet" to get the bulldozer/piledriver CPU working up to potential. well to have a CPU (or any hardware) dependent on an OS (or any software) to be able to perform . . just smells too much like an apple to me.
(close can of worms)
now on a positive note, there is one thing that an AMD B/P chip will do phenomenal (pardon the pun); it will make a very excellent F@H rig!
eddieroolz
Splendid
- Sep 6, 2008
- 7,057
- 0
- 25,860
rds1220
Splendid
- Sep 10, 2011
- 3,717
- 8
- 23,815
[citation][nom]looniam[/nom]how many more attempts will it take to admit reality?people can talk about patches until they are blue in the face but what matters is results. and so far the outcome has fallen terribly short of the speculated, desired or hoped to achieve results.[/citation]
Yes exactly right. Software isn't going to magically make AMD's crappy micro architecture amazing. Like DJ said no amount of software has no chance of fixing that. All the problems are hardware restriction.
Yes exactly right. Software isn't going to magically make AMD's crappy micro architecture amazing. Like DJ said no amount of software has no chance of fixing that. All the problems are hardware restriction.
Just wait! Steamroller will be out in no time to fix Bulldozer / Pile drivers performance issues with a 1-2% performance increase!!! Go AMD!
/end sarcasm
I'm just going to point out here that I was right, yet again, from the start. You can't design something "for the future" in a market that needs something for today. Multithreading is nice and dandy, but it will never ever replace strong single thread performance, we live in a serial world. And again, as I said before, I heard the same rumors back in the late 90's about the world changing to multithreaded OS's and programs in the early 2000's. It didn't happen then, and it isn't happening now. Don't expect it anytime soon.
If anything, Multiprocessing control needs to be taken away from the OS and put back into the hardware, where it use to be. The less the OS has to deal with, the better. Single core, multiple ALU'S and FPU's is where it's at. It simplifies programming by taking that difficult task and letting the hardware deal with it instead of the programmer, plus it removes the major overhead multiple cores create in a multithreaded OS.
Sorry, still pissed off at how far technology has gone off the tracks.
/end sarcasm
I'm just going to point out here that I was right, yet again, from the start. You can't design something "for the future" in a market that needs something for today. Multithreading is nice and dandy, but it will never ever replace strong single thread performance, we live in a serial world. And again, as I said before, I heard the same rumors back in the late 90's about the world changing to multithreaded OS's and programs in the early 2000's. It didn't happen then, and it isn't happening now. Don't expect it anytime soon.
If anything, Multiprocessing control needs to be taken away from the OS and put back into the hardware, where it use to be. The less the OS has to deal with, the better. Single core, multiple ALU'S and FPU's is where it's at. It simplifies programming by taking that difficult task and letting the hardware deal with it instead of the programmer, plus it removes the major overhead multiple cores create in a multithreaded OS.
Sorry, still pissed off at how far technology has gone off the tracks.
For those who are bemoaning the lack of PCI-E 3.0 on AMD --erm, unlike Intel's Z77 mainstream chipsets, which are limited to 1 x16 or 2 x8 in SLI/CF (natively), the 990FX can have 2 full bandwidth x16 lanes for its "ancient" PCI-E 2.0 slots, so the end result is exactly the same. (in general x16 PCI-E 3.0 bandwidth is about equal to x32 PCI-E 2.0). The only modern intel chipset that can boast of full x16 multiple slots natively is X79 (Sandy Bridge E/forthcoming Ivy Bridge E).
blazorthon
Glorious
- Sep 24, 2010
- 14,022
- 21
- 43,265
[citation][nom]BigMack70[/nom]They should... AMD CPUs (unless running at about 10 GHz) are going to bottleneck a multi-GPU setup, badly in some games...I guess at triple-screen resolutions maybe not, but on any single monitor, yeah...Proof? I've seen nothing that suggests a single card benefits from PCI-e 3.0 x16 over PCI-e 2.0 x16, though I haven't looked into this particular claim.[/citation]
Actually, there are some non-gaming applications that can have noticeably improved performance with PCIe 3.0 x16 over PCIe 2.0 x16, but generally not greatly so and most applications do not benefit from it. I could pull up benchmark comparisons showing this, but I don't know any that aren't outdated.
One big thing to keep in mind is that such work would mostly be done on workstations/servers and any such system that only uses PCIe x8 instead of x16 (very common in past systems, but IDK how common it is right now) would have a significantly higher performance boost in such applications than if it had PCIe 2.0/3.0 x16 connectivity.
Actually, there are some non-gaming applications that can have noticeably improved performance with PCIe 3.0 x16 over PCIe 2.0 x16, but generally not greatly so and most applications do not benefit from it. I could pull up benchmark comparisons showing this, but I don't know any that aren't outdated.
One big thing to keep in mind is that such work would mostly be done on workstations/servers and any such system that only uses PCIe x8 instead of x16 (very common in past systems, but IDK how common it is right now) would have a significantly higher performance boost in such applications than if it had PCIe 2.0/3.0 x16 connectivity.
blazorthon
Glorious
- Sep 24, 2010
- 14,022
- 21
- 43,265
[citation][nom]classzero[/nom]I would like to see if the ARM can out perform AMD BD[/citation]
You'd have to be more specific than that. If you want to compare *ultra mobile* ARM CPUs to desktop Bulldozer CPUs, AMD would still win by huge margins. There are neither ultra mobile Bulldozer CPUs nor desktop ARM CPUs right now, so we can't do an apples to apples comparison. Maybe ARM will be able to outperform Bulldozer's current top CPUs like ten years from now, but otherwise, they almost definitely couldn't do it, maybe unless they scale up to match desktop/laptop CPU power consumption and die area levels.
You'd have to be more specific than that. If you want to compare *ultra mobile* ARM CPUs to desktop Bulldozer CPUs, AMD would still win by huge margins. There are neither ultra mobile Bulldozer CPUs nor desktop ARM CPUs right now, so we can't do an apples to apples comparison. Maybe ARM will be able to outperform Bulldozer's current top CPUs like ten years from now, but otherwise, they almost definitely couldn't do it, maybe unless they scale up to match desktop/laptop CPU power consumption and die area levels.
- Status
TRENDING THREADS
-
-
-
News TSMC founder says Intel should focus on AI, not advanced process technologies- Started by Admin
- Replies: 27
-
Discussion What's your favourite video game you've been playing?- Started by amdfangirl
- Replies: 4K
-
News Intel CEO Pat Gelsinger retires, effective immediately — also steps down from BOD, two co-CEOs step in
- Started by Admin
- Replies: 124
Latest posts
-
-
-
-
-
Question Why does my PC have a black screen when I launch my games ?
- Latest: Edward Paradero
Facebook
Twitter
Instagram