10 Most Recently Updated Pages

RadeonHD: Mobility Chipsets now Supported

Hans de Ruiter — Mon, 28 Feb 2011 15:16:35 -0600

With the Sam460ex recently going on sale and including the RadeonHD driver, the driver has inevitably been tested on a wider range of hardware. While most hardware is working just fine, two people with a Sapphire Radeon HD 4350 reported (see here and here) that it didn't work with their cards. Most of their screens were filled with garbage. This surprised me as all Radeon HD 4350s tested previously worked just fine. So, suppressing my desire to grumble about people ignoring my recommendations (hey, it wouldn't have happened if you had got a Radeon HD 4650/4670 like I suggested ;-) ), I got to work tracking down the problem.

A quick look at the offending card's PCI product ID revealed that it was actually a Mobility Radeon M92 chipset on the board, and not an RV710, which is what Radeon HD 4350s should have. This explained why all Radeon HD 4350s tested to date worked just fine, while these failed miserably. While this was a clear differentiating factor, it did not explain why it failed.

These cards highlighted two problems, not one. First, there was the obvious problem of the cards not working. However, the driver should have detected the problem, and disabled hardware graphics acceleration, so that the display was readable at least. The cause of both problems lay in the GPU initialisation code. This code checks which chipset the card has, and then uploads the correct microcode and executes the appropriate initialisation. As it turns out, some chipsets - most notably all of the mobility chipsets - were missing, and so no microcode or initialisation was done. A small bug prevented this failure from being detected, and so hardware acceleration was not disabled.

The good news is that I have fixed these issues, and future versions will be able to use these cards. In the mean-time, I strongly encourage people to not buy Sapphire Radeon HD 4350 cards, as some of them will not work with the publicly released driver. Likewise, I discovered that the current driver will not initialise RV740 and RV790 chipsets, so avoid the Radeon HD 4770 and 4890 cards too. I do not expect people to buy these cards for their Sam460ex's, as they will block the PCI port. Nevertheless, you have been warned.

RadeonHD Card Recommendations for AmigaOS 4.x

Fri, 14 Jan 2011 17:48:56 -0600

Many people have been asking me for advice as to which Radeon HD graphics card they should buy for their Amiga. So, here are my recommendations. For the lazy, there is a quick summary at the end of the page. But first, a quick disclaimer:

Disclaimer: These recommendations are my personal opinion based on my current knowledge and experience with these cards. They are not based on objective testing; I have not tried and tested every card. As such I make no claims as to the accuracy of these recommendations. Moreover, I could change my mind later based on new data. Ultimately, only you can decide which card you think us best for your situation.

At present the RadeonHD.chip driver supports any card from the Radeon X1300 to the Radeon HD 4890. While the driver has not been tested with every single card available in that range; to date, all cards have worked. If you are worried that certain cards may not work, check the compatibility page for a list of tested cards. There is no support for the newer Radeon HD 5000 and 6000 series yet, so please do not get one of those.

These recommendations are made from the viewpoint of what will be best once both the 2D and 3D drivers have been completed. While the current drivers (as on Saturday 15 January, 2011) lack compositing support for Radeon HD 2000-4000 series cards; this will not remain so for long. When I started writing the drivers my goal was to bring support for modern graphics cards complete with shader support to the Amiga, and it will happen. There are no release dates yet, but it is coming. Anyone who is nervous about buying before the drivers are finished are going to wait anyway.

PCI-Express Capable Amigas

My first word of advice for those with a brand new PCI-Express (PCIe) capable Amiga, is to forget about the low-end cards, unless you get one for almost nothing. The reason is that low-end really does mean just that: low -end. They have slower Video RAM (VRAM), lower clock-speeds and fewer stream processing units (or shader units in the Radeon X1000 series). For example, a Radeon HD 4890 has 800 stream processors running at 850 MHz and a VRAM bandwidth of 124.8 GiB/s, whereas a Radeon HD 4350 has 80 stream processors running at 575 MHz and the VRAM bandwidth is 8 GiB/s. This makes a big difference. A Radeon HD 3870 has 320 cores at 725 MHz and VRAM bandwidth of 57.6-72 GiB/s, and it will leave a Radeon HD 4350 in the dust. In fact, a Radeon X1950 pro will beat the Radeon HD 4350 at simple operations such as 2D blits, because it has greater VRAM bandwidth.

That is not to say that the Radeon HD 4350 is a bad card; because that would be wrong. After all, it has greater capabilities than older generation cards. However, if you have spent money on a new PCIe capable Amiga, then get a graphics card with performance to match. A mid-range to high-end card would make much more sense.

So, which card would be best? My suggestion would be to choose a card from the Radeon HD 4000 series. Sure, the 2000 and 3000 series will work, but why bother? You are unlikely to save money, and the 4000 series have more capabilities. Likewise, the Radeon X1000 series will work just fine. However, they are getting harder to find, and they have more limitations. The only reason that I can think of to use an older card is because you happen to have a spare one lying around.

The next question is which specific model within the range. Well, that depends entirely on what you are looking for, and how much you are willing to pay. For example, you may be like me, and want high performance, but not the noise that typically comes with it. The first GLSL capable card that I ever used was the Radeon X800 that I used for my university research. At the time it was the most powerful graphics card ever made. It also sounded like a rumbling jet when running under load. I like quieter machines, particularly if they are located in my bedroom. I have found the HIS Radeon HD 4850 iCooler IV, and the Sapphire Radeon HD 4890 Vapor-X. Both of these are quieter than the other models with active cooling. For absolute silence, the highest performance card that I could find was the Gigabyte Radeon HD4850 1 GB GV-R485MC-1GI, which is passively cooled. The next step down is the Sapphire Radeon HD 4650 Ultimate Edition, which is a passively cooled card. They did make a Radeon HD 4670 Ultimate Edition too, but it is no longer available. There is also the HIS Radeon HD 4650 iSilence. The passively cooled Gigabyte card is rather hard to find though. Fortunately, it is possible to make your own passively cooled card using the Arctic Cooling Accelero S1 Rev 2 VGA Cooler (it has been done).

IMPORTANT: While the passively cooled Gigabyte has been tested with no air-flow going over the cooler, and will survive, it is important that the air in the case is well circulated with a passive high-end card. If you are unsure of how to do this properly, then get a card with a fan.

What about VRAM? Generally speaking, more is better, but more VRAM also uses more power (e.g., GDDR5 has a reputation for being power hungry). Having said that, even 512 MiB is significantly more than existing Amiga OS 4.x users with Radeon cards are used to. Unlike the older Radeon cards, these cards do not have the split device problem that prevented most 256 MiB card owners from using the full 256 MiB. The drivers will eventually be able to use all of the available VRAM. However, there is one slight issue. Picasso96 assumes that all VRAM is accessible via the CPU, but this assumption no longer applies. Only the first 256 MiB on these cards is actually directly accessible and the rest is only directly accessible by the GPU. This was most likely done to avoid running out of addressing space (2 GiB of VRAM would only leave 2 GiB for all other devices and RAM on a 32-bit system). Regardless, as a consequence Picasso96 will only be able to use 256 MiB. I can almost hear people saying, "but I thought that you said..." Yes, the rest of the VRAM will still be used. The 3D drivers will be able to make full use of the additional VRAM, and I am investigating additional possible uses. The current drivers (as on Saturday 15 January 2011) already use the additional VRAM as a temporary buffer for overlapped blits.

AmigaOne-X1000

If you have one of these machines, congratulations on getting the most powerful Amiga to-date. The AmigaOne-X1000 (A1-X1000) will come as a complete machine, and will already have a suitable graphics card.

Sam 460ex

When selecting a card for the SAM 460ex, it is important that you check the thickness of the card. The reason for this is that many cards take up two slots, and the PCI bus is right next to the 4x PCIe bus (which uses a 16x slot so that graphics cards fit). If you have any interest in using the PCI slot - and believe me, you probably do want to use that slot - then you will need a card that takes up a single slot. This rules out the high-end cards, as all of those (as far as I know) have a big cooler attached to the top that takes up an additional slot. Likewise, some of the passively cooled cards have a double-height heat sync. My Radeon HD 4350 PCI card is like that. Many of the mid-range cards take up only one slot, although not all of them. I have personally got a Sapphire Radeon HD 4650 Ultimate Edition, which fits nicely. I can also confirm that the Radeon X1950 pro (at least some of them), also fit nicely, if you prefer one of those. The best thing to do is to check the card's size before you buy.

The Sapphire Radeon HD 4650 Ultimate Edition's passive heat-sync wraps around the board, so it does not interfere with the next slot

PCI Only Amigas

While I would recommend that people upgrade to one of the newer motherboards that support PCIe, I realise that many will want to upgrade their existing machines instead. My current recommendation is still that people wait until the driver has actually been released before buying a card for older Amigas. There is no set release date yet.

With older AmigaOS 4.x machines, the choice is rather more limited. The performance will be limited by the PCI bus' bandwidth, and the lower CPU speed. Nevertheless, I still see upgrading to a Radeon HD (or Radeon X1000 series) card as a step up. Not in terms of speed, but in terms of capability. These cards will introduce the ability to use shaders, which opens up a whole new set of possibilities. Here are some general things to keep in mind:

WARNING: Do not try to use an AGP version of these cards in your Amiga! Despite having made it clear that this is not possible, I still see people talking about it occasionally on various web forums (or fora, whichever you prefer). I want to make it clear right now that this will not work, and will likely damage both your graphics card and your Amiga. The reason is that the Radeon HD cards are all AGP 4x/8x cards, and do not support the higher voltage of the AGP 2x slot. The AmigaOne XE mistakenly has a universal AGP connector, so the graphics card will physically fit. Please do not try this
If you use a PCI-to-PCIe bridge card, get one with a PEX8112. The Radeon HD driver has code to optimise the settings of the PEX 8111 and 8112, but not other PCI-to-PCIe bridges. Plus, it supports blind-prefetching, which increases VRAM read access speed. Also, you will need a low-profile graphics card, or your card will be dangling over the edge, with connectors blocked by your case
A Radeon HD 4650 plus a PCI-to-PCIe bridge card will outperform a Radeon HD 4350 PCI. This is a case of my original assumption being wrong. Originally I said that, due to the slow PCI bus and lower CPU power, the Radeon HD 4650 would be a waste. Well, someone decided to try anyway and, it does have a noticeable improvement over the Radeon HD 4350 PCI. Yes, the PCI bus and lower CPU power is a bottleneck, and you'll never be able to get the maximum out of the card. Nevertheless, testing has shown that it does make for a snappier machine overall

SAM 440-Flex and SAM 440ep

REMINDER: The RadeonHD driver is not yet available for these motherboards.

At this stage I would recommend that Sam 440-flex users to get either a Radeon HD 4350 PCI, or get a PEX 8112 based PCI-to-PCIe bridge card and a low-profile Radeon HD 4650** if they wish to upgrade to a Radeon HD card. It is faster than older cards, and offers all of the capabilities of the Radeon HD 4000 series. If you want to get a Radeon X1000 series card, then the Radeon X1550 has the best performance.

** NOTE: I say "low-profile" because a PCI-to-PCIe plus a low-profile Radeon HD card should be able to fit into a full-size slot, whereas plugging a full-size

AmigaOne-XE (without the DMA Fix) and the AmigaOne-SE

REMINDER: The RadeonHD driver is not yet available for these motherboards.

My recommendations for these motherboards are the same as for the Sam 440-flex.

AmigaOne-XE (With DMA Fix) and the Micro AmigaOne

REMINDER: The RadeonHD driver is not yet available for these motherboards.

It is well known that the Micro AmigaOne (uA1) has trouble with using a plug-in graphics card. Some have managed to use plug-in graphics cards, but had to sacrifice DMA for their hard-drives, and USB. AmigaOne-XE (A1-XE) boards with the DMA fix have the same problem as the uA1 with graphics cards plugged into the 33 MHz bus. Unfortunately UBoot does not initialise Radeon HD cards properly when plugged into the 66 MHz PCI slot since it cannot handle nested PCI bridges. I hope that the AmigaOne UBoot update project manages to fix this issue. In the meantime, know that using Radeon HD cards (or Radeon X1000 series cards) will require sacrificing DMA for IDE and USB.

Pegasos-II

IMPORTANT: Radeon HD cards will not work with the Pegasos-II.

The Pegasos-II's openfirmware will not detect and initialise the graphics card. It will quite happily detect and initialise the PCI-to-PCIe bridge on the card, but it will not detect the actual graphics chipset that resides behind the bridge. Attempts have been made to contact Genesi and B-Plan regarding getting the firmware updated, but to no avail.

Classic Amigas with PowerPC Accelerators

IMPORTANT: Radeon HD cards do not work in Classic Amigas.

Apart from being impractical, PCI-to-PCI bridges are not supported on any of the Classic Amiga bus-boards. Since connecting a Radeon HD card to a PCI bus requires a PCI-to-PCIe bridge (either soldered on to the card, or separate), this will not work. The bridge chip is detected, but not the graphics card behind it. Even if this problem were solved, I have been told that the 256 MiB of CPU-accessible VRAM into the Zorro address space (NOTE: The 256 MiB must also be 256 MiB aligned, due to PCI alignment rules).

To be honest, the Classic Amiga is so underpowered relative to these cards that using them together makes little sense. You really need a modern PCI-Express capable Amiga to be able to take full advantage of Radeon HD cards.

A Quick Summary

If your Amiga has a PCIe slot, then it is recommended to choose a mid to high-end Radeon HD 4000 series card
When choosing a card for the Sam 460ex, be mindful that the PCI slot is right next to the 4x PCIe slot. Choose a card that takes up only a single slot, and does not have an oversized heat-sync/fan that would block use of the PCI adjacent slot
A1-XE motherboards with the DMA fix and uA1 motherboards require disabling DMA for the IDE drives and USB in order to use any PCI graphics card (in the 33 MHz slot)
Radeon HD cards (and Radeon X1000 cards) do not work in the Pegasos-II
These graphics do not work in Classic Amigas
The drivers will eventually be able to use the full amount of VRAM that these cards have
The ultimate choice in card is up to you. Whatever you choose, enjoy

GfxBench2D the First Release

Hans de Ruiter — Sat, 18 Jun 2011 20:26:03 -0500

Welcome to yet another development log. This post marks the first release of the GfxBench2D benchmark tool, and the corresponding web application (web-app). The benchmark tool tests the performance of graphics cards for a range of different operations, and (optionally) uploads the results to this website for display. While it is still a work-in-progress (isn't most software?), it has reached a point at which it is useful.

The original idea for this project was a result of my Radeon HD driver for AmigaOS 4.x project. Graphics card driver writers are particularly interested in how fast the graphics card can perform operations. So, I started writing a 2D graphics benchmarking tool so that I could measure performance, identify any areas that need work, and monitor progress as the driver is developed further. This helped me track down the PCI-to-PCI bridge initialisation problem that was seriously hampering performance on Sam440 machines (you will find details about this here, on the last paragraph in the section called "The Long Road to Here").

While the early version of the benchmarking tool was helpful, collating results and generating graphs manually was rather tedious and time consuming. I wanted to get results from more graphics cards, and others were asking for benchmark results too. So, I thought about how I could automate the process and enable other people to send results quickly, and the GfxBench2D web-app was born. A web-API was developed to upload results, and an upload client was added to the benchmark tool. At this point I also added the tests that I still wanted it to perform.

As mentioned earlier, this is still a work-in-progress. In particular, there are more features that I would like to add to the website for displaying and comparing results. It is also planned to port the benchmarking tool to other operating systems too. In the meantime, I invite you to check out the results that have already been submitted, and/or download the benchmarking tool to try out on your own machines (AmigaOS 4.x only at the time of writing).

GfxBench2D for Windows

Hans de Ruiter — Thu, 18 Aug 2011 03:25:01 -0500

I am please to announce that the Windows version of GfxBench2D is now available. Windows XP or higher is required. The website code has also been updated in order to accommodate multiple Operating Systems (OSes).

If any developers are interested in helping to create a version of GfxBench2D for other OSes, please contact me.

GfxBench2D updates for AmigaOS and Windows

Hans de Ruiter — Mon, 22 Aug 2011 03:06:58 -0500

Both the AmigaOS and Windows versions of GfxBench2D have been updated, and are available for download. Some bugs were found in the Windows version over the last few days, resulting in version 2.2. At this stage the Windows version is still considered to be beta. The AmigaOS version is now at 2.1. It incorporates some bug-fixes and changes that were primarily the result of developing the Windows version.

Update 2011/08/23: The Windows version is now at 2.3 after another minor bug was found.

GfxBench2D

Wed, 01 Jun 2011 18:31:36 -0500

Welcome to the GfxBench2D results database. GfxBench2D is a benchmark test tool that measures the 2D performance of graphics cards. Download the GfxBench2D software to test your own graphics card(s). Visit the Forum to discuss benchmarks, make suggestions, etc.

The tests are designed to look not only at how fast graphics cards are, but also why (i.e., which operations are fast, and which are slow). This is achieved by measuring the speed of the most common operations at various sizes.

GfxBench2D Download

Wed, 01 Jun 2011 19:23:46 -0500

GfxBench2D is a benchmark test tool that measures the 2D performance of graphics cards. Test results from various computers and graphics cards are available online.

Please download the version of GfxBench2D for your operating system:

AmigaOS 4.x - Download (view results)
Windows (XP or higher) - Download (view results)

Once downloaded, simply install the software and run it. There are only a few simple prompts to follow. Uploading results to the online database is optional, but recommended; the more results are uploaded the more useful the information is to everybody.

Thank you for trying out GfxBench2D.

NOTE: Ports of GfxBench2D for other operating systems are planned. If a version for your operating system is not available yet, please be patient. Since I cannot support all operating systems myself, I do plan on setting up a test server so that those willing to help with porting, can do so. However, this is not up and running yet. If you are interested to help out, then please contact me via email.

Source Code

I will be releasing the source-code at some point. However, I still have to decide in what form. If there is sufficient interest from others to help porting then I may set up a Google code project. Otherwise it will be provided as a downloadable archive here. If you are interested to help out, then please contact me via email.

GfxBench2D for Windows Updated to Version 2.4

Hans de Ruiter — Sat, 03 Sep 2011 19:41:54 -0500

GfxBench2D for Windows is now at version 2.4. It is available at the usual place, i.e., the download page. This version fixes a few bugs, the most important of which is a bug that prevented certain multi-core machines from running the tests. If you find any bugs, please contact me via email, or post to the GfxBench2D forum.

GfxBench2D for Windows 2.6

Hans de Ruiter — Mon, 10 Oct 2011 02:09:18 -0500

The Windows version of GfxBench2D has received another update. This version includes a workaround for an issue with buggy graphics drivers. On a small number of computers it would hang at the WritePixelArray test. This was due to a driver bug affecting the CopySubresource() function. It would allow the source-buffer to be locked for writing before copying of data from that buffer to the graphics card had completed, potentially enabling the old data to be overwritten before it had been used. Fortunately, there was a different method to wait until the operation was done, which waits for the GPU to be idle. This same method is used prior to each test to ensure the most accurate results.

The updated version can be found at the usual place.

Adbrite Dumped for Poor Performance

Hans de Ruiter — Tue, 27 Apr 2010 00:00:00 -0500

It's been a while since I have posted anything to this blog, mainly because I have been busy with other things. I have, however, still kept an eye on it, deleting spam when necessary, and monitoring how the advertising programs are performing. Today I made the decision to dump Adbrite.

A few months ago, Abrite revenue dropped to almost nothing (over a 10 times drop), and stayed there. I decided to give it a few months before deciding whether to continue using their service or not. Well, revenue has remained very low, and a pitiful few impressions are actually shown. Adbrite advertisements also had a poor click-through rate due to poor relevance. Hence, today I ditched that advertising network.

Ironically, round the time that the sudden unexplained revenue drop occured, I was about to ditch ADSDAQ for providing mostly ugly slimming and teeth whitening advertisements, and stick with Adbrite. Since then, has ADSDAQ improved dramatically, and filtering has removed the offending advertisements.

Acording to this article, someone else had a similar dramatic drop in revenue from Adbrite after they limited his accound for invalid clicks without any notification. This may have happened in my case, because the month prior to the revenue collapse my account recorded a few clicks (yes, performance was that poor). I am not going to bother checking if this is the case because I would still dump Adbrite for the following reasons:

Failing to notify customers/partners of important events (such as restricting their account for alleged violations of terms and services) is completely unacceptable business practise to me, and
If my account has been restricted for alleged "invalid clicks" which I did not make or cause, that would demonstrate a very poor invalid click detection algorithm on which I cannot rely, whether I am advertising, or providing advertising space.

So, either Adbrite has suddenly started performing very poorly, or their "invalid clicks" technology and business practises are unacceptable. Either way, they're gone.

I still haven't found an advertising network that I am satisfied with. ADSDAQ is okay, but has a low fill-rate (impressions per page view). Chitika works quite well, but caters only to visitors from the USA and Canada. Google Adsense was okay, until they disabled my account due to invalid click activity from an unknown source (I still have no idea who was responsible). I have just started testing Brinkin Banner Exchange, which combines concepts from both banner exchanges, and contextual advertising. It is an interesting concept; we shall see how it performs.

NOTE: Brinkin suddenly just disappeared a long while ago, and so I can only assume that it didn't make enough money in order to survive. (2012/03/9)