Announcement

Collapse
No announcement yet.

NVIDIA GeForce GTX 280M

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • NVIDIA GeForce GTX 280M

    Phoronix: NVIDIA GeForce GTX 280M

    After launching the GeForce 200 series last year, NVIDIA unveiled the GeForce GTX 260M and 280M GPUs for notebook computers earlier this year. The GeForce GTX 280M is currently NVIDIA's fastest notebook GPU, even though it is derived from the GeForce 9800GTX+ core rather than the GTX 280 desktop variant. This 55nm notebook GPU has 128 processing cores, supports two-way SLI, features NVIDIA PureVideo HD technology (important for VDPAU usage under Linux), and other features to pack a desktop performance punch in notebook computers. The Linux-friendly System76 manufacturer recently introduced the Bonobo Professional notebook computer with the NVIDIA GeForce GTX 280M and an Intel Core 2 Quad processor, which we happen to be looking at now and today are delivering some initial performance results from this high-end NVIDIA GPU under Ubuntu Linux.

    http://www.phoronix.com/vr.php?view=13969

  • #2
    yeah, another rebrand of years old hardware to rip off users. Well done nvidia!

    Btw, is the mass dying problem finally solved?

    Comment


    • #3
      @energyman: I almost feel like asking you to stop trolling around and stay in opensource radeon forums , although I fully understand your frustration as an ATI customer. In any case, only the GTS250 (and below) is a rebranded 9800 (which is a rebranded 8800), but the GTX260M and GTX280M are indeed new. And that mass dying thing - it was not mass, else nvidia would have gone bankrupt, and it was with few 8000 series cards - though I understand it was quite a few cards and it was annoying. And FYI, I have a notebook with 8400M GS (Dell Vostro), which has been running fine for almost 24x7 for about 2 years between 48-56 in normal work conditions, including some heavy gaming (and I used to overclock till I used to hit 100 celsius sometimes) during early days. In general, I have a feeling that discrete graphics require some good ventilation - macbook pros are good, but are very tight for fitting a class-A/B gpu - so much that they are often crippled by Apple in hardware/software itself - like the mobility raden x1600's sometime ago.

      Meawhile, the tests here are good, but could be better - the card is way better than it is shown - it is perhaps the top of the line mobile graphics (except for multi-gpu setups). What is notably missing is the GTKPerf tests - on my Asus G50VT with 9800M GS, the entire test for 1000 rounds is gunned down within 70 seconds. (And such a test would silent all of the few gtk-pervs around who complain about slow 2d.) I am not sure about how much Nexuiz is optimized (it is a great game btw), but benchmarking a few commercial games like Doom3 or ET:QW wouldn't have hurt. (Meawhile, me waiting for Wolfenstien on August 2009 .)
      Last edited by hdas; 06-19-2009, 12:49 PM.

      Comment


      • #4
        For those interested in Unigine Tropics, Unigine Sanctuary, or Enemy Territory: Quake Wars results for this GPU under Linux, using the 185.18.14 display driver there was a bug preventing OpenGL from being initialized in those tests.
        Micheal, are you seeing this on all nv cards? I'm wondering because I've been benching those fine with the 185.18.14 drivers and has no problems initializing openGL. Just for example: http://global.phoronix-test-suite.co...09-15605-12784

        Comment


        • #5
          what trolling? Nvidia is renaming their old crap for the second time now - ripping of people. That is not trolling, that is a FACT.

          Nvidia based laptops are also mass dying out there. Dell, HP, Apple, everybody has the same problem. Again FACT not trolling.

          So, before you even DARE to accuse me of trolling, check the facts. Just because you are lucky does not change anything.

          It is a mass dying. Nvidia has two class actions against it and they are sued by their insurrance for lying.

          You don't believe me? Fine, go to the inquirer and check for yourself.

          Comment


          • #6
            Yeah Inquirer and Wikepedia and so on . If you faced something bad on yourself, please do tell us - or better, go to the appropriate place. What you don't mention is the action the OEMs and Nvidia took. For example, when the problem arose, I was worried too about my Dell Vostro. Within a short time, Dell did extend the warranty to two years - no excuses. Likewise, I am sure Apple did their part, and hopefully the rest of the manufacturers and so did Nvidia.

            Meanwhile, what makes you think that GTX280M is a rebranded one?

            Comment


            • #7
              Originally posted by hdas View Post
              Meanwhile, what makes you think that GTX280M is a rebranded one?
              All of the new products that were released is basically rebranded old crap.

              Comment


              • #8
                Sorry, I take my words back regarding the rebranding thing. Indeed GTX280M is apparently a rebranded *desktop* 9800GTX . I was assuming it was related to the desktop GTX280, which is new. What is a little disappointing is that the bus is still 256-bit (my 9800M GS has that too, except that 280M has double the shaders).

                PS: I am happy with the old crap .

                Comment


                • #9
                  Originally posted by hdas View Post
                  Yeah Inquirer and Wikepedia and so on . If you faced something bad on yourself, please do tell us - or better, go to the appropriate place. What you don't mention is the action the OEMs and Nvidia took. For example, when the problem arose, I was worried too about my Dell Vostro. Within a short time, Dell did extend the warranty to two years - no excuses. Likewise, I am sure Apple did their part, and hopefully the rest of the manufacturers and so did Nvidia.

                  Meanwhile, what makes you think that GTX280M is a rebranded one?

                  nvidia took no actions at all. They lied for more than a year about the problem, then put 100m aside - not very much, if you look deeper into the problems. The numbers are in the open for everybody to check. From SEC filings, to filings by the insurance company and the forums of almost everybody who ever sold a nvidia laptop.

                  But continue to live into your little fanboy bubble if you prefer....

                  Comment


                  • #10
                    Originally posted by hdas View Post
                    Sorry, I take my words back regarding the rebranding thing. Indeed GTX280M is apparently a rebranded *desktop* 9800GTX . I was assuming it was related to the desktop GTX280, which is new. What is a little disappointing is that the bus is still 256-bit (my 9800M GS has that too, except that 280M has double the shaders).

                    PS: I am happy with the old crap .
                    Ya it's the same chip as the desktop 9800GTX but it's not quite the same as when they rebranded chips like the 8800GT to a 9800GT. This is at least the first time this particular configuration of G92b that it has been featured on a mobile platform.

                    Comment


                    • #11
                      Originally posted by phoronix View Post
                      While one can easily play an HD video with an Intel Core 2 Quad Q9000 and not experience a bottlenecked CPU no matter the video output, as can be seen from the graph above, when using VDPAU there is virtually no CPU usage at all.
                      I am sorry, but that is very wrong. You do not specify what movie you used to test "1080p playback" but it obviously is a very simple, low-bitrate clip. Most other 1080p content is a lot more demanding on the processor and a 2Ghz Core would be hard pressed to play it smoothly, let alone at less than 100%!

                      The reasons for that are simple:
                      1. None of the Free/OSS media players support multithreading, so the 4,8,or 64 cores are wasted. This is mostly caused by ffmpeg which still, after all these years, does not do multi-threaded decoding.
                      2. Seeing as only one core is used, the speed becomes very important! I've ran a lot of tests with various video files, and the conclusion is that one needs at least 3Ghz for proper playback of most content. Even then there will be some movies that have trouble in certain high-movement parts, for example when there is snow, rain, or flocks of birds flying around.

                      As a proper test, I recommend two clips:
                      1. The Matrix Reloaded intro, with the "green/digital" clock -- that will run sluggish @2Ghz but should be ok @3Ghz.
                      2. Scenes with millions of flying birds from "Planet Earth". Those become a slideshow even at 3Ghz.

                      Of course, a lot of other content will play fine, and the slowdown will only be observed in high-action scenes, but that doesn't make a 2Ghz processor fit for the job, at most it's barely adequate. I would also like to see benchmarks with VDPAU for a movie that brings the processor to it's knees, otherwise these low-bitrate tests are quite useless...

                      Comment


                      • #12
                        really? I have an option in xine
                        'FFmpeg video decoding thread count'

                        and could you point to some video material that is
                        a) sanely&good coded
                        b) 1080p

                        that can be legally downloaded somewhere?

                        Thank you.

                        Comment


                        • #13
                          Originally posted by mgc8 View Post
                          2. Scenes with millions of flying birds from "Planet Earth". Those become a slideshow even at 3Ghz.
                          Otherwise known as the killa sampla, found easily with google. (Which btw plays fine on even a lowly 8200 IGP with vdpau).

                          Comment


                          • #14
                            oh, and why is there a 'threads' useflag for ffmpeg if it doesn't support threads?

                            maybe your distribution ships a crappy package?

                            Comment


                            • #15
                              Originally posted by energyman View Post
                              oh, and why is there a 'threads' useflag for ffmpeg if it doesn't support threads?

                              maybe your distribution ships a crappy package?
                              Ummm, for encoding perhaps? There is a reason why there is a separate project called ffmpeg-mt. (Which is what you use if you want to utilize the threading setting in xine)

                              Perhaps you missed this article a while back.

                              http://www.phoronix.com/scan.php?pag...nterview&num=2

                              Diego Biurrun felt that this multi-threaded version of FFmpeg may still be years away.
                              Last edited by deanjo; 06-20-2009, 02:31 PM.

                              Comment

                              Working...
                              X