Dual-head DVI ViewFinder graphics card announced

A dual-head DVI-capable ViewFinder graphics card for RiscPCs and A7000s was announced today by hardware guru John Kortink. This new kit, which connects off-the-shelf PC hardware to legacy Acorn machines, allows punters to hook two DVI-quality monitors to their RISC OS computers.

Links

Previous: NetSurf on RISC OS faces axe
Next: Prototype Acorn Unix-powered A680 lands on eBay

Discussion

€99 is a bit steep for such an ancient card, but I suppose he may have had to jump through some hoops to source such archaic technology. Is it known what card is being used, so people have the option of purchasing elsewhere? rjek on 14/8/09 7:15PM [ Reply | Permalink | Report ]

To me 99 euros is a very affordable price to get a DVI output, and users of existing View Finders who could have considered buying the Vpod will stay 'Kortink Inside'. No doubt RISC OS 6 drivers will soon be available. Xavier on 15/8/09 10:01PM [ Reply | Permalink | Report ] Affordable does not equal not expensive or over-priced. And I was under the impression that RISC OS 6 still lacked any support for Radeon-based ViewFinders. rjek on 16/8/09 12:00AM [ Reply | Permalink | Report ] Compared to the vPod, I find John's latest offering extremely non-expensive and surely not over-priced, seeing that the hardware alone will easily cost 30€. And as we all know, software development for a small market is expensive. The missing RO6 drivers are a problem of course, but have ViewFinder users really "up"graded to RO6? After all, most of the cool ViewFinder features are no longer available there. So I guess it is more of a problem for ROL, who still have not properly documented the ViewFinder driver's state of RO6 on their website. hubersn on 16/8/09 12:30AM [ Reply | Permalink | Report ] Perhaps I was confused by the article saying the driver software was made freely available, and the €30 card was being sold for €99... Of course, this dual-DVI enabled ViewFinder is way more expensive than a V-Pod, once you include the cost of the VF and the card you have to throw away to fit the new one rjek on 16/8/09 11:42AM [ Reply | Permalink | Report ] [link] I'm lost ... Xavier on 16/8/09 7:36AM [ Reply | Permalink | Report ] That dictionary isn't very good. Don't you think something can still be expensive, and yet within your budget and thus afforded? rjek on 16/8/09 11:43AM [ Reply | Permalink | Report ] I believe it's a cheap upgrade for existing owners of View Finder cards...for them it's more affordable to spend about 85 GBP than 149 GBP for the Vpod. But well if you've got some other ideas it doesn't really bother me. I've ordered an upgrade for J.K. and will be very happy to get an excellent quality picture thanks to the DVI output. I just can't believe the VPOD offers VGA output only in year 2009. Even if it's faster, I vote for quality of image 1st, within a budget. Xavier on 16/8/09 4:56PM [ Reply | Permalink | Report ]

Rather fantastic news for Viewfinder card users. Hope the software is RISC OS 6 compatible. sa110 on 14/8/09 7:47PM [ Reply | Permalink | Report ]

Pre RISC OS Six only. It's a Radeon AGP card. egel on 14/8/09 8:08PM [ Reply | Permalink | Report ] According to John's mailing list posting, the software is not RISC OS Six compatible. It's a Radeon card, and I think this is completely unsupported on RISC OS Six. hubersn on 14/8/09 8:09PM [ Reply | Permalink | Report ]

It's fantastic to see RiscOS's most tallented developer(s) continue to support the platform - well done! I guess £99.00 & lack of Select support are issues but, hey, it's a niche platform regarding the former and the latter would be fixable (with sufficent positive encouragement). Many thanks JK, I'll at least give some serious consideration to upgrading mine. Charlie on 15/8/09 11:51AM [ Reply | Permalink | Report ]

I suppose what all the viewfinder users need to do now who want to run both the new card and RO6 is to email developer@riscos.com and advise them of this. The more who make their wishes know, the more chance of it happening. sa110 on 15/8/09 11:54AM [ Reply | Permalink | Report ] I wouldn't hold your breath, they refuse to even fix their driver for the ViewFinder cards it currently supports. On RISC OS 4.X with JK's firmware I used 2048x1536x32bpp@70Hz for 5 years, with RISC OS 6 it insists the card is only capable of 1920x1440x32bpp. If 2048x1536 is selected it only does 16bpp at a completely useless 48Hz. druck on 19/8/09 10:08PM [ Reply | Permalink | Report ] It's not 99 GBP but 99 euros so you save about 17%, bringing the price down to about 82 GBP. It's why I find it very affordable (cheap, inexpensive, as read in a dictionary) , considering it brings dual DVI output to the RISC PC, a feature not offered by the Vpod, sadly. Xavier on 16/8/09 7:49AM [ Reply | Permalink | Report ] Oopsss, webmaster please can you correct my comments ? I wanted to write but and euros and forgot the '/' in the tags ... Thanks. Xavier on 16/8/09 10:08AM [ Reply | Permalink | Report ]

Anyone aware of which card its being upgraded to? Would there be any advantage to be had in moving on from a Rage 128 card, for performance I mean? I don't have two monitors or a DVI input so that's no pull. Monty on 16/8/09 11:24AM [ Reply | Permalink | Report ]

And I understand the V-Pod trounces the VF if you're using RO6. I assume the bottleneck with the VF is sending commands and data over the podule bus, so I doubt putting a card in with faster 2D operations is going to give you a noticeable benefit. rjek on 16/8/09 11:46AM [ Reply | Permalink | Report ] Problem is I have a 1920 X 1200 screen. I understand the VPod can't do that at 16M colours. Otherwise, yes, I suppose that would be the better and faster option. What you're saying though, if I understand rightly, is that there isn't likely to be any single screen benifit from the new card. Monty on 16/8/09 12:11PM [ Reply | Permalink | Report ] ?? Fantastic picture quality thanks to DVI is a major benefit in high resolution mode. Xavier on 16/8/09 5:00PM [ Reply | Permalink | Report ] Surely this doesn't apply to monitors which only take vga (or hdmi) input? Or is the the higher quality some how retained even though its converted to vga? Monty on 19/8/09 1:36PM [ Reply | Permalink | Report ] Agreed - performance-wise AIUI VPod > VF (R128) using RO6 driver > VF (R128) on 4.X using JK driver, due to the bus bottleneck and native video abstraction support on RO6 respectively. The actual GPU speed is unlikely to have a significant impact AIUI, simply because on the Rage 128 on the RPC even when commands are queued there is often insufficient data for a 'command queue' to occur - graphics operations are performed 'instantly' as far as the OS is concerned! md0u80c9 on 16/8/09 6:40PM [ Reply | Permalink | Report ] Are there any meaningful benchmarks out there that show the speed advantage of the VPod? I only tried it out briefly and could not detect any meaningful speed difference between my VF-equipped RiscPC and the VPod-equipped machine. Anyway, lack of DVI is a definitive no-no. hubersn on 16/8/09 7:44PM [ Reply | Permalink | Report ] I do have a fillrate test, here : [link] This will give a good idea of the general speed of rectangular drawing operations, which are by far the most important in the desktop. jk on 20/8/09 8:45PM [ Reply | Permalink | Report ] On an A9home at 1024x768 @16 million colours. This reports a file rate of 596 Mb/sec. Not sure if this is good or bad. sa110 on 20/8/09 9:52PM [ Reply | Permalink | Report ] On a Risc PC running RISC OS 4.39 and a Radeon ViewFinder it is between 1200 and 1500 MB/sec. jk on 20/8/09 10:40PM [ Reply | Permalink | Report ] IYONIX pc at 1600x1200 in 16M colours: 1900MB/sec using the old graphics card from the first batches. hzn on 10/9/09 9:23AM [ Reply | Permalink | Report ] Something a bit off with the maths there, unless you are using a 260Hz refresh rate. I've always run 2048x1536x32bpp at 75Hz on my Iyonix with both old and new graphics cards. The overall video transfer rate is 900MB/s, but it outputs data at 1157MB/s during the display part of each line. druck on 10/9/09 12:58PM [ Reply | Permalink | Report ] Unless I've misunderstood, they're talking about the "general speed of rectangular drawing operations", rather than the transfer rate from the board to the screen, which is what you calculation is of (I think). So, the boards can shift around several screens' worth of pixels around the screen per frame. Stoppers on 10/9/09 8:55PM [ Reply | Permalink | Report ]

I don't know where this nonsense about acceleration performance comes from. Radeon has a memory bandwidth of 1.2 GB/sec. SM501 has half that. Both have similar command overheads and both are behind the podule bus. So unless the ViewFinder driver of RISC OS 6 is absolutely shoddy compared to the VPod's (and even not then, since for operations where acceleration is most relevant, overhead is least so), ViewFinder will generally be much faster than VPod. jk on 19/8/09 12:55PM [ Reply | Permalink | Report ]

Surely instead they will both be limited to the same speed; that of the rather ancient motherboard? Monty on 19/8/09 1:31PM [ Reply | Permalink | Report ] You don't realise exactly how acceleration works. It involves sending a parmeterised command to a drawing engine, which then does the drawing all by itself. The podule bus is only involved in the transfer of the command (which is, say, 8 words long). I.e. the drawing speed is completely dominated by the memory bandwidth available to the drawing engine (e.g. 1.2 GB/sec in Radeon's case), not by the bandwidth of the podule bus (8 MB/sec with DMA). This is true for all drawing operations that benfit most from acceleration, i.e. where the drawing engine draws a lot of pixels, like rectangle copy (used for moving windows around the desktop) and rectangle fill (used for clearing window and menu backgrounds). In ViewFinder's case, since it caches sprites in the video card's memory as well, it even applies to sprite plotting (if the sprite data was not cached, it would have to pass the podule bus on each draw). jk on 19/8/09 1:55PM [ Reply | Permalink | Report ] So only when using the RISC OS 6 video abstraction interface a larger part of the drawing operations are acceleratable, can a V-pod be faster, compared with a viewfinder under RISC OS 4? That could be possible, but why would a viewfinder under RISC OS 6 be slower than a V-pod? Inefficient driver or does the RISC OS 6 video abstraction interface uses many drawing operations that can only be accelerated on a SM500/SM501? egel on 20/8/09 12:35PM [ Reply | Permalink | Report ] It doesn't use an SM500, either I strongly suspect that the ATI card has more acceleration features. I'm beginning to suspect it might have something to do with how the SM50x's and the ATI's memory and registers are mapped into podule space, and how things are shuffled around when there. rjek on 20/8/09 1:40PM [ Reply | Permalink | Report ] The V-Pod doesn't use an SM501. Without having the source to either driver, we won't know. All I know is lots of people say the V-Pod is faster. rjek on 19/8/09 6:57PM [ Reply | Permalink | Report ] The bottleneck is the podule bus, not the internal memory interface of the GPU: I suspect the difference between the cards lies in efficiency of memory transfer over the podule bus, and the relative complexity of setting up accelerated commands on the R128 versus SM50X chipsets. Adding a faster GPU will help just about as much as adding a shiny new SSD drive to the RPC's interface - you will still get <6MB/sec data transfer, even though the drive could manage nearer 300MB/sec! NB both drivers were written by the same author and have the same featureset (indeed share the same common codebase). Most of the tests I'd seen were done by the author and are common, reproducible operations - complex draw file rendering, circle drawing, desktop redraw time, accelerated sprite plots, etc. It would be relatively easy for someone to independently write some test programs in BASIC to test this. md0u80c9 on 19/8/09 10:31PM [ Reply | Permalink | Report ] The podule bus is only the bottleneck for operations that hardly benefit from acceleration in the first place (e.g. line drawing). These operations have little or no impact on overall desktop performance. If you would expend the effort to make them as fast as possible anyway, it's simply a matter of limiting the amount of data going across the podule bus. There's no difference in potential for that between ATI chipsets and SM50X. Their programming models are virtually identical. jk on 19/8/09 11:36PM [ Reply | Permalink | Report ] You are possibly the only person who can answer this, but perhaps a big difference is the way they are bridged to the podule bus? The SM50x can be run as a PCI device, or a memory-mapped device; I imagine the logic in the CPLDs is much simpler. rjek on 20/8/09 12:20AM [ Reply | Permalink | Report ] Think I've realised why you might see this differently. On OS 4 you cannot easily queue the commands. On OS 6, command queueing is used on both R128 and VPod (because of the use of data aborts to stall and unstall the pipeline when direct memory access occurs), which is why the OS 6 performance is faster for all cards. In isolation, I agree you'd see little difference. But when you queue them together, polyhline operations are significantly faster than their direct memory access fill equivalents, and yet the podule bus is still the bottleneck here. As Rob's pointed out, podule bus logic may also come into play. In either case, the numbers will speak for themselves - would anyone care to write a quick BASIC app or two to test graphics drawing ops on the different drivers? md0u80c9 on 20/8/09 9:51AM [ Reply | Permalink | Report ]

Hmm wonder if we can persuade John to talk PCI for this? [link] epistaxsis on 19/8/09 5:07PM [ Reply | Permalink | Report ]

Stop drooling over video cards and read your emails, there's show organising to be done 43 days to go... [link] helpful on 20/8/09 5:34PM [ Reply | Permalink | Report ]

A question if anyone 'in the know' is still looking @ this thread: RadeonVF + RO6 is a 'no-go' from everything I've read. As I understand it this is becuase of a lack of Radeon drivers in RO6 - shame, I've finally had a chance to try out RO6 & rather like it but not enough to ditch my VF (especially as I'm seriously considering taking JK up on his upgrade)... ...Now, my understanding is that JK's drivers aren't available as a 'default' because RO6 is soft-loaded so they get 'kicked-out' when the OS loads. Would one potential 'fix' be to soft-load JK's drivers (modules) early in !Boot after RO6 has started? Or is it a case of the RO6 GFX sub-system being too differnt for the 'standard' RadeonVF driver to work even if present..? My thanks to any who can shed light on this - I'd like both but the VF is more important to me. Charlie on 24/8/09 9:33AM [ Reply | Permalink | Report ]

No. JK's driver AIUI works by patching the OS's graphics routines directly. The ROL driver uses the abstraction interface provided. The video layer in RO6 works quite differently and the code is in totally different places. The reason that OS6 kills the original VF driver on startup is simply that if the original driver manages to initialise on RO6, it stiffs the machine completely. Of course this happens way before you get to the *command prompt, so it kind of renders the machine unusable . I'm doubtful of the benefits for the workload involved of updating the R128 driver (given one exists) nor for the Mach64 (given its obsolescence), but I'm sure John could write a very nice video driver for RO6 for the Radeon if there was interest - with the added benefits discussed earlier in the thread . md0u80c9 on 24/8/09 2:06PM [ Reply | Permalink | Report ] Thanks for the info, just a thought for a way of getting the combination working. I think (as suggested above) I'll e-mail ROL & politely request they consider support for the RadeonVF... ...As you suggest JK is possibly in the best position to tackle this, assuming he has time. I'll make the same request of him when I call about the above upgrade. Idle curiosity: Are the Radeon versions rare? Everyone I know with a VF has a Radeon-based one. Charlie on 24/8/09 2:36PM [ Reply | Permalink | Report ] Significantly rarer than the R128 version AFAIK (JK would be the definitive source of course!) IIRC, John released a Radeon variant shortly before he stopped supporting the ViewFinder - so logically there shouldn't be many of these around at all! Each card (Mach64/R128/Radeon) works totally differently to the others, and unhelpfully many different varieties of the same card (eg. different manufacturers of the Rage 128) exist and behave differently too. This is probably why the ROL driver supports only a subset of cards (most R128s) - development time for three drivers and their different card varieties becomes quite prohibitive. Mach64 support is pointless - a quick trip to EBay or a RISC OS dealer should be able to pick you up a R128 AGP card for very little. An Ebay search just now suggests purchase prices of £5 for these cards. Caveat emptor - note that not all varieties will work, but for that price you can afford to buy an unsupported variety once or twice; RISC OS dealers will no doubt charge more but at least guarantee finding a working variant. Some people also get utterly confused as to which card is in the machine (assuming R stands for Radeon, or 'It's ATi so it must be a Radeon'), which doesn't help! Generally Mach64s had 8MB RAM, Rage 128s had 32MB RAM, and the Radeon 9200 which I think John supplied at one time has 128MB RAM. I think on the original driver *ViewFinderInfo told you more about the card. Hope that all helps. md0u80c9 on 24/8/09 3:04PM [ Reply | Permalink | Report ] Practically all of the differences between AGP cards of the same class (Mach 64, Rage 128, Radeon) are in their initialisation. This is probably still done by the ViewFinder firmware before RISC OS 6 silences it. The RISC OS 6 driver for ViewFinder, therefore, hardly needs to deal with differences between AGP cards of the same class. Mostly only with acceleration. And even between classes, there are very few differences in implementing acceleration. jk on 24/8/09 3:40PM [ Reply | Permalink | Report ] Many people have upgraded to the Radeon version, because it offers a lot more features (Dual Head, TV out, DVI). It would have been a lot better to provide a Radeon driver for RO6 instead of a R128 driver, because upgrading to Radeon is easy and cheap. However, bringing people to accept a big downgrade when they upgrade their OS is a difficult position. hubersn on 24/8/09 3:46PM [ Reply | Permalink | Report ]

I thing we at CJE may have sold to end users more than 50% of all viewfinders sold. We have also supplied quite a few AGP cards as upgrades. Fewer than 10% were the 128MB version and 90% of the the upgrades have been 8 to 32MB so I'm confident that not much more than 10% of ViewFinder users are using 128MB [1]. AIUI JK software is almost pure magic! Well its patching of the OS had to be quite hairy Then came along select, this complicated things significantly, it was explained to me by one of the Select programmers that it lets the VF software patch the OS and then went back and repatched the OS to suit, very messy. WIth RO6 abstracting the video driver patching patches became impractable so they had to build in support for specific cards. JK obviously has a lot of experience of differences necessary for different cards, ROL much less so. I suspect it would be quite a lot of work for JK to change to using the official RO6 video abstraction. On the other hand I'm not sure if the relevant ROL programmers are able to divert the necessary time and finance [2] to add support. [1] Continental users In our experience tend to go for the top of the range in our experience so I can believe that many/most of the users Steffen knows about are 128MB. [2] ROL would have to try and obtain/be loaned most variants of Radeon 128MB that have been used and given the variation in output options there must be at least 3 and possible quite a few more! CJE on 25/8/09 12:32PM [ Reply | Permalink | Report ]

ROL would have to try and obtain/be loaned most variants of Radeon 128MB that have been used and given the variation in output options there must be at least 3 and possible quite a few more! Still, not impossible. Having already support built in for one VF variant, adding another variant or two should not be beyond their capabilities, and whilst not generate any more income, would help to maintain the current subscriber base. Afterall, what is the point of subscribing to Select if you cannot use it! sa110 on 25/8/09 12:44PM [ Reply | Permalink | Report ] The core role of Select is OS development, not hardware support/driver development. ViewFinder previously had to be supported as an exception because no developer was available to maintain a then-obsolete but highly valued podule and OS changes forced a compromise position. A side-effect to the OS is the development and progression of the graphics abstraction interface, which has been extended over the last few Select releases (and now covers more functions than covered by the original 4.XX driver). This week's activities suggest VF is now in active development - and new hardware/driver developments are usually logically the responsibility of their author, just as would be the case with any other new hardware. Given that John is actually developing the software here, not the hardware, clearly his development time is significant and reflected in the upgrade price. Suitable Radeons can be purchased for pretty decent prices, so John's predominant charge here is for changes to his driver. RO6 is in no way broken by the release of a Radeon driver for VF under 4.XX - the previous options of the VPod or R128 ViewFinder still exist. ROL run a business - so assume /all/ work has to at least break even for it to be worthwhile. Which either means to support a new feature either prices go up, or other features aren't developed. Would you therefore accept, say, an equivalent 50 euros price hike to Select to cover additional driver development costs (allowing the purchase of cards at 49 euros)? This simply reflects the upgrade price John is presumably putting on his time for equivalent development. Or instead, should other OS development be slashed back to cover it, and non-Radeon owning RO6 users (ie. all current subscribers) be paying for an update they don't currently use, and in the case of some users (VRPC) will never use? Or, should it be left to the original developer as to whether he wishes to create Radeon support on RO6 at a later date? Not suggesting a right answer here - just pointing out the straightforward problem of finite resources and development rationing. md0u80c9 on 25/8/09 10:35PM [ Reply | Permalink | Report ] Given the choice of a working ViewFinder and the RISC OS 6 changes, it's ViewFinder everytime for me, the Risc PC is unusable without it. I ditched my entire RPC because of RO6's poor support for the Rage 128 cards, if I'd had a Radeon I would never even contemplated wasting money on RO6. druck on 26/8/09 7:27AM [ Reply | Permalink | Report ] But the new icons! They're so lovely! rjek on 26/8/09 8:20PM [ Reply | Permalink | Report ] I think the bigger economic interest in solving the incomplete compatibility of RISC OS Six with ViewFinder lies with ROL. ViewFinder is not back in production, and there's a very limited number of upgrades. Once these are gone, it's sort of back to blissful anonymity. Dual-head DVI is mainly something I wanted myself, and since I'm probably not the only one, I decided to make it available. jk on 27/8/09 12:05AM [ Reply | Permalink | Report ]

'AIUI JK software is almost pure magic! Well its patching of the OS had to be quite hairy' FWIW: A while ago I pointed a friend in the direction of JK's website for interest. He has an amazing ability to hack unexpected hardware into old computers - sadly he's not into Acorns. Having seen what JK did and examined the files available he was full of praise for what he regarded as a 'proper bit of hackery'. I certainly found a new appreciation for the work that must have gone into this GFX upgrade... Charlie on 25/8/09 1:29PM [ Reply | Permalink | Report ]

A brand new card is on sale on Ebay .... I hope it means I'm going to get my already paid upgrade this week ... Xavier on 30/8/09 7:49PM [ Reply | Permalink | Report ]

Auction link: [link] The seller is a one jhjkortink. diomus on 30/8/09 8:00PM [ Reply | Permalink | Report ]

Confirmed. All ordered upgrades will ship this week. Great news. Xavier on 31/8/09 8:07AM [ Reply | Permalink | Report ]

ViewFinder software version 1.75 is now available via : [link] Simply follow the instructions. From 'Updates' : - Bugfix : monitor's maximum linerate setting (as flashed) was never effective, since it wasn't observed correctly. - Bugfix : the flash utility sometimes (if only rarely) failed to program the 29C512 due to an undiscovered timing sensitivity. - Added support for dual-head DVI output. The specifications of the second DVI capable monitor can be specified via the flash utility. - In dual-head setups, the second head no longer needs to display the same vertical resolution as the first (i.e. the vertical resolution can now be specified via *ViewFinderMode). - Added '*Configure ViewFinderDVIScaling' option, which selects the monitor (0) or the AGP card (1) to perform scaling (to the monitor's native resolution). The AGP card cannot scale the second head, so 0 is the recommended setting for dual-head DVI operation. - Changed dual-head mode related syntax of *ViewFinderMode. egel on 31/8/09 5:47PM [ Reply | Permalink | Report ]

I installed my Dual Head upgrade yesterday, it's amazing to have a 44" RISC OS desktop!! JK, you've done an amazing job! robheaton on 5/9/09 9:06AM [ Reply | Permalink | Report ]

Got mine as well, will fit it at the weekend hopefully. Pete on 10/9/09 10:24AM [ Reply | Permalink | Report ]

Please login before posting a comment. Use the form on the right to do so or create a free account.

+++ Message board +++

Read messages or start a new thread

Search the archives

Today's featured article

Online banking with RISC OS
We thought this would be of high interest
 28 comments, latest by devine on 25/9/05 8:30AM. Published: 1 Sep 2005

Random article

ROL surveys Select32 interest
You can't miss it
 56 comments, latest by TonyStill on 23/11/04 7:04PM. Published: 15 Nov 2004

Login

Useful links