Page 2 of 5 FirstFirst 1234 ... LastLast
Results 11 to 20 of 48

Thread: New hardware needed

  1. #11
    Quote Originally Posted by Lifepower View Post
    Drivers for AMD GPUs have issues almost on all platforms, so for stability I'd suggest Nvidia.
    What isues? I'm hardcore AND/ATI supporter and in all theese yers I recall to have problems ony 3 times:
    1. Back on my ATI Radeon 9200 I downloaded once corupt driver package (old driver and new Catalyst Control Center). ATI has fixed the problem in one day.
    2. On my old laptom I once had some problems becouse I actually instaled wrong graphics drivers (mobile drivers also contain information about Flat Panel Screen). So my picture looked a bit garbled (Flat panel was still conected with graphics card throug VGA conection). In todays laptops Flat Screens are conected to graphics card through Display port which alows drivers to get the necessary information from the panel itself.
    3. Last problem I had aprox. a mont ago when I intentionally downloaded beta drivers and they contained a bug which prevented OpengGL to work properly (some artifacts in 3D games), DirectX renderning worked OK.

  2. #12
    Quote Originally Posted by SilverWarior View Post
    What isues? I'm hardcore AND/ATI supporter and in all theese yers I recall to have problems ony 3 times:
    1. Back on my ATI Radeon 9200 I downloaded once corupt driver package (old driver and new Catalyst Control Center). ATI has fixed the problem in one day.
    2. On my old laptom I once had some problems becouse I actually instaled wrong graphics drivers (mobile drivers also contain information about Flat Panel Screen). So my picture looked a bit garbled (Flat panel was still conected with graphics card throug VGA conection). In todays laptops Flat Screens are conected to graphics card through Display port which alows drivers to get the necessary information from the panel itself.
    3. Last problem I had aprox. a mont ago when I intentionally downloaded beta drivers and they contained a bug which prevented OpengGL to work properly (some artifacts in 3D games), DirectX renderning worked OK.
    What about 64bit Linux and AMD drivers? Any experience with that?
    Best regards,
    Cybermonkey

  3. #13
    PGD Staff code_glitch's Avatar
    Join Date
    Oct 2009
    Location
    UK (England, the bigger bit)
    Posts
    933
    Blog Entries
    45
    Okay, Im seeing some ATI/AMD slamming going on here... Lets clarify:
    ATI driver 'issues' usually revolve around a few things:
    On windows, if you upgrade to the latest catalyst and do a full uninstall/reboot/reinstall windows driver in device manager - you might get slapped a little. Eg: a game may crash when some look nice feature is enabled. Have I experienced this? No. Why? When ATI said, ooh look we just made CCC 12.8 and released it 6 hours ago I waited a day, typed 'catalyst 12.8 issue' into google and found that it was quite stable, so I upgraded. This will avoid you 99% of problems. Also, if your setup of catalyst is fine - in other words nothing wrong with it: don't touch it for funzies and go to the absolute bleeding edge. Just like with open source. If you do - you sort of deserve a bit of pain in exchange for being on the bleeding edge.

    On linux things are not as clear cut: you can opt for open/closed source drivers.
    open source radeon drivers: IMHO is nicer - it has almost all the chinks ironed out now, runs great for your desktop, is native as they come and is built right in. The down sides are that it can hurt DirectX performance under wine by some 10-40%+ depending on the game. Although this is to be expected somewhat - does linux used DirectX? Not really. So why include it? There you go.
    Closed source fglrx: if you go this route you get the ATI speeds on everything albeit ATIs driver team has its reputation. In 3 years of having these installed I cam accross 1 issue which is shared with nvidia cards: docky has a random bug in it and one from my ignorance as a result of a badly botched Xorg.conf for multi-monitor support which had some 'garbling' in some 3d opengl textures.. wiping Xorg.conf and getting an updated deb for the game fixed it so I can't complain.

    So what it comes down to is whether you NEED the very best, macish driver support, albeit usually at an extra cost. Since this is not a 600 euro build, AMDs offerings tend to have the market well in hand.

    SSDs: I've had my SSDs 2 years, its my primary and only disk - it has 158.2 days of power on time in 340 power cycles. SMART data says not a single sector is bad, and thats the same experience for anyone in my family who has one for work as well as family friends (although one of my friends' intel drive failed after 18 months...) I would conclude that if your drives die after 2 years you're dong something wrong in some way. For example: SWAP file/partition. Bad BIOS/driver/firmware/application setup... Or any number of factors. In short its inconcievable for an SSD to have largely inferior life to an HDD - it has no moving parts, it has management to keep sectors equally used and it does things in a timely manner which avoids users getting annoyed and going into buzzkill mode on the mouse and thrashing the disk.

    As much as I like intel - they sit in all my laptops at the moment and an old desktop from the sandy bridge announced and about to ship hype era... Well, old.... Its a dual core deal at some crazy GhZ.... Intel run hot. Very hot. This was blamed on north/south,xyz bridges and chipsets and then intel put them on die and it was the magic pixies' fault as to why a die shrunk, locked down Xeon runs hotter than anything seen before on a custom cooler is beyond me. My X3 is oced on its stock cooler by a fair margin - and its rarely hits 50C. GPU wise my 5750 hits 85 at peak times. Thats not a big deal - 100-105C is usually the overclockers start being careful, 110C is the problem zone and 115C is the emergency shutoff and 120-125C is the damage zone.

    Thats my opinion - and although it sounds like AMD fanboyism I have 1 AMD machine and my sister has one. The other 5 are intel machines and I'll leave it at that. True AMD doesn't do the highest end stuff, and if I were serving up databases for datacentres a dual Xeon/i7 it would be...
    I once tried to change the world. But they wouldn't give me the source code. Damned evil cunning.

  4. #14
    Quote Originally Posted by code_glitch View Post
    Okay, Im seeing some ATI/AMD slamming going on here...
    Not at all, I've stated that although it's raw power is excellent, the drivers are might be less polished than those in Nvidia, which can also be troublesome, especially if Optimus is used. I've actually suggested all three brands - Intel (for HD 3000), Nvidia (more polished, less power) and ATI (more power, less polished), while you seem to have special bias towards AMD.

    Quote Originally Posted by code_glitch View Post
    On windows...
    With my Asus HD 6780 and XFX HD 6770 in the office, I've been using most driver versions from 12.x series, including 12.8 on all fresh installs. Neither allowed to turn on/off VSync and multisampling in Direct3D 9. If you enable VSync in Catalyst, it doesn't work - you have to use D3DOverrider. Enabling multisampling in Catalyst for random app will force it on any other application. If you don't use any of these features, or somewhat complex GLSL shaders in OpenGL in Linux/Mac OS X, then there's nothing to worry about.

    Quote Originally Posted by code_glitch View Post
    I would conclude that if your drives die after 2 years you're dong something wrong in some way.
    Please don't be upset for contradicting your recommendations. There is no need to start another [Overclock, AMD GPUs, SSDs] vs [Not-overlock, other GPUs, HDDs] religious war. As for the reasoning I've quoted, I'm sure you understand why it is a logical fallacy.

    P.S. Our drives failed after 2-3 months of use, not 2 years.


    Quote Originally Posted by code_glitch View Post
    For example: SWAP file/partition. Bad BIOS/driver/firmware/application setup...
    For the record, we've disabled SWAP file on SSD and used certified Asus motherboards with latest BIOS updates. Why our SSDs died out quicker? Probably because of usage patterns as we've been recompiling and linking projects many times daily, not to mention full 20-30 Gb project rebuilds.

    You also didn't mention that lifespan of SSD might be considerably lower than of HDD under certain usage patterns due to limited write cycles. However, if we don't mention write cycles, many problems in SSD are due to bugs in firmware logic. For practical applications, SSDs should not fail under any of usage scenarios unless write cycles become a problem (and even those are mediated by using some random relocation). That's why I suggested using SSDs being as playing a roulette - in your case they did not fail (yet), but in many others they do, so the question remains: do the benefit of improved hard disk seek time outweigh the additional purchase cost and the risks of sudden SDD failure along with full data loss? This is something OP needs to take into account for purchase. You've put SSDs as perfect choice, albeit expensive. I've added increased risks to the equation.

    Yes, calculating such risks is not easy, but with HDD you have warning signs (e.g. mechanical wear) and as long as plates are undamaged you can still recover your data, while in SSDs with fauly logic chip you may get data corruption even before you experience complete failure.
    Last edited by LP; 11-09-2012 at 05:46 PM.

  5. #15
    Hey guys, I don't wanted to start a flame war here. As for SSD: they are no option for me now. I am using HDDs since more than 15 years and never had any issues, so why change?
    Best regards,
    Cybermonkey

  6. #16
    Quote Originally Posted by code_glitch View Post
    Intel run hot. Very hot. This was blamed on north/south,xyz bridges and chipsets and then intel put them on die and it was the magic pixies' fault as to why a die shrunk, locked down Xeon runs hotter than anything seen before on a custom cooler is beyond me.
    Well, at least in AMD FX 8150 that I've mentioned, the temperature increase is slow as you would expect, where the CPU case is heated first, then the room and so on. However, our Core i7 rigs have temperature spikes under load from 30-35C to 60C and then slowly increasing to 70C. I thought temperature transfer was a problem, but after touching headsink tubes - they are burning hot and can actually melt your fingers! The only OC I have is for RAM speed, but the difference between my rig and my partner's is only a couple of degrees. There is a funny thing out of this: when it's cold, I'm using CPU stress test to turn workstation into room heater.

  7. #17
    PGD Staff code_glitch's Avatar
    Join Date
    Oct 2009
    Location
    UK (England, the bigger bit)
    Posts
    933
    Blog Entries
    45
    2-3month failures? wha? I mean. Thats impressive, you guys should stress test for manufacturers SSD wise my recommendation is based on the end users experience of it feeling fast. We can go theoretical all day long. Eg: what in gods name was AMD doing when they came up with bulldozer and why intel does what it does... (this links with the final point...)

    I can definitely agree ATI drivers aren't the best in that going in with a wrecking ball and leaving that as is might be an improvment in some cases... But credit where its due - ATI drivers are very compatible... I recall using some nvidia code for a GTX card on my 5750 to get some experimental stuff working with OpenCL. I mean seriously - nvidia GTX drivers for an ATI card. Well played Though I'm impressed that it doesn't affect ATI users more than just knowing what each checkbox does and whatnot.

    Didn't mean to be agressive, sort of skim read your reply as I'd just got home and caught onto ATI, Driver, glitches, intel, intel, i7 and... Yeah. Rule of thumb - FX series tend to be innovative and new. With that comes rather a lot of 'stuff' that we simply can't quite get our heads around. Eg. why they use less power run a little hotter and deliver less power in some scenarios and much more in others with no relation. Personally I'm avoiding intel anything - since the AMD switch was my revelation that 60C stock is NOT acceptable compared to 40C on an unlocked and OCed chip with stock coolers. As for graphics, I like ATIs value though I'm a harsh customer when it comes to drivers...

    My bias towards AMD is more of a 'okay intel does everything you will ever do for 20 years at 500 frames/sec at double the cost'. Okay. But why? You really wont notice the 440frames that dont get displayed on your monitor. But I'll notice the $200 alright XD And to top it off, intel have had VERY aggressive marketing in the UK and people buy intel because its 'cool' and has 'HD graphics' with 'enhanced yadida' and so on. Then theres the ultrabook bribery and whatnot which tends to use the market to attain a monopoly. I have a similar attitude on the sainsburys vs tesco wars. Of course I wont sustain that a Phenom ii x6 or top of the line AMD is the best for everybody - I recognise intels' xeons trump opteron, i7 trump phenoms and etcetera. I however dont agree with intel promoting cpu power alone and using marketing to convince everybody it is fact. If intel have their way - developers can take the cyanide pill: they only just got OpenGL 3.1 more or less right and their latest support for anything is not quite up to what I call an ATI driver kludge.

    Though on the heat issue, its worth noting: ATI heatsinks while gaming. No touching. Ever
    I once tried to change the world. But they wouldn't give me the source code. Damned evil cunning.

  8. #18
    Quote Originally Posted by Cybermonkey View Post
    What about 64bit Linux and AMD drivers? Any experience with that?
    No! I must admit that I have almost no expirience with Linux. So pardon me on this.

  9. #19
    Quote Originally Posted by code_glitch View Post
    2-3month failures? wha? I mean. Thats impressive, you guys should stress test for manufacturers SSD wise my recommendation is based on the end users experience of it feeling fast.
    From what I read around the web the biggest killers for SSD-s are:
    1. Windows Defrag which does a lot of data moving back and forth. Fragmentation barely slows down SSD drives especialy those with TRIM support so defragmenting files on them is almost pointles. And Windows XP has disk defragmentation enabled on all drives by default. Windows Vista does have defragmentation enabled on all drives by default if the computer has atleast one standard HDD (might have been fixed with SP1). Windows 7 is suposingly smart enough to keep defragmentation active only for normal HDD-s.
    2. Windows Indexing service which is part of Windows Serch 4 can also cause quite some stress on SSD-s due to frequent reading of varios files which are being indexed especially when there is indepth indexing enabled (indexing contents of supported files).
    3. Windows tends to constantly acces lots of log files, writing whole bunch of data in them. Not to mention how often it does reads from varios system files. Download process monitor to see how much disk activity there is even when Windows is idle. You will be surpized.
    You can get Proces Monitor from: http://technet.microsoft.com/en-us/s...rnals/bb896645
    4. Using uTorrent with SSD can also be a real killer. The reason for this is the fact that uTorrent always operates with small chunks of files. Not to mention the fact that it rapidly reads and writes data from his Chache folder. And if this folder is on SSD it will cause quite a lot of stress on it.

    Quote Originally Posted by code_glitch View Post
    I can definitely agree ATI drivers aren't the best in that going in with a wrecking ball and leaving that as is might be an improvment in some cases... But credit where its due - ATI drivers are very compatible... I recall using some nvidia code for a GTX card on my 5750 to get some experimental stuff working with OpenCL. I mean seriously - nvidia GTX drivers for an ATI card.
    Probably it all worked becouse ATI graphics cards have very good support for OpenCL while there are still some newer nVidia graphics cad which don't support OpenCL.
    And NO nVidia Cuda is not the same as OpenCL. OpenCL is the emerging standard which tries to unify approach for using varios technologies like nVidia Cuda, AMD Stream or now caled APP, etc.

  10. #20
    Quote Originally Posted by code_glitch View Post
    2-3month failures? wha? I mean. Thats impressive, you guys should stress test for manufacturers SSD wise my recommendation is based on the end users experience of it feeling fast.
    We purchased SSDs exactly to speed up compilation process, which improved 30 minute compilation time to 20 minutes when doing full rebuild of 20 Gb project. This would typically involve processing (reads, some writes) of more than 100,000 files. If there is a bug in SSD logic chip, this will make it fail rather quickly. I agree that it's not a typical end-user usage scenario, but honestly for me it doesn't matter if it will fail now or at some random point later - I would rather use hardware that I know is quite reliable, in which case it is HDD (I just stay away from Hitachi and Samsung/Seagate brands).

    Quote Originally Posted by code_glitch View Post
    Didn't mean to be agressive, sort of skim read your reply as I'd just got home and caught onto ATI, Driver, glitches, intel, intel, i7 and... Yeah. Rule of thumb - FX series tend to be innovative and new. With that comes rather a lot of 'stuff' that we simply can't quite get our heads around. Eg. why they use less power run a little hotter and deliver less power in some scenarios and much more in others with no relation.
    Well, despite online reviews, AMD FX 8-core series are pretty powerful, at least for us where we use them for work and not necessarily multi-threaded. Yes, in our rigs Core i7 2600K beats AMD FX 8150 in single-threaded applications, but when things become multithreaded, they both work very well.

    Quote Originally Posted by code_glitch View Post
    Personally I'm avoiding intel anything - since the AMD switch was my revelation that 60C stock is NOT acceptable compared to 40C on an unlocked and OCed chip with stock coolers.
    That is because AMD's stock coolers at least have some cooper in them, while Intel's coolers are made of some poor quality aluminum mixed with sawdust. I once tried to cut off piece of aluminum from Intel's cooler (to make a gasket as a replacement for laptop's deep thermal pillow) only to see it crumble into pieces under pressure.

    Quote Originally Posted by code_glitch View Post
    they only just got OpenGL 3.1 more or less right and their latest support for anything is not quite up to what I call an ATI driver kludge.
    On that note, I was wondering. Does Apple ever update their graphics drivers on OS X? Because Intel HD 3000/4000 with latest Windows drivers handles OpenGL 3 on Windows very well, while on Mac OS its OpenGL 3 support is buggy and detrimal.

Page 2 of 5 FirstFirst 1234 ... LastLast

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •