I'm sorry, I have more important things to do on a Sunday so just a quick fire response..

The extra cache on K6-2+ means it can setup triangles multiple times faster than the K6-2. It also means it can spin through the orchestrating software layers (e.g. Direct3D) faster. That brute force is going to boost FPS on the K6-2+ regardless of PCI/AGP bus.

I had a 3Dfx Banshee SGRAM, so 3Dfx definitely put SGRAM on AGP cards too. It's not a PCI/AGP differentiator, but it shows that test cards need to be carefully hand picked before running PCI/AGP comparisons.

In the later PCI-via-southbridge vs 8xAGP-3.0 comparison, not only is the PCI slot severely handicapped by being architecturally demoted to the chipset periphery, but the actual cards compared a 64-bit VRAM PCI card vs 128-bit VRAM AGP card. As a PCI/AGP comparison it's totally ludicrous!

MattRocks wrote on Today, 09:06:

The extra cache on K6-2+ means it can setup triangles multiple times faster than the K6-2.

lets get real, K6-2+ is still slower than Intels 1998 low end of a Celeron 300A. Nothing runs 'multiple times faster' on it compared to K6-2 😀 even if you gimp it by completely disabling L2 and loading motherboard with 512MB of uncached ram https://www.philscomputerlab.com/amd-k6-2-vs- … -vs-k6-iii.html
Besides two other examples use proper configs with real boy processors 😀

MattRocks wrote on Yesterday, 21:48:

There's nothing wrong with fake AGP because real AGP was a gimmick. Or, more cynically, AGP was Intel's attempt to tame the grap […]
Show full quote

PcBytes wrote on 2026-06-04, 18:18:

ChrisK wrote on 2026-06-04, 11:58:

The ASROCK board I think doesn't have a real AGP port. It's different from the 775i65G(no V) variant.
But not sure atm. Just want to say have an eye on that.

Purple AGP = fake AGP. That's how ASRock got AGP on non-AGP chipsets (845GV, 865GV), using their "AGI" branded slot.
https://theretroweb.com/motherboards/s/asrock … 5i65gv-rev-1-00

There's nothing wrong with fake AGP because real AGP was a gimmick. Or, more cynically, AGP was Intel's attempt to tame the graphics card industry.

Think about it. The year was 1997. Top end monitors were boasting resolutions of 1800×1440. To render basic desktop GDI at 1800x1440 you need >600MB/s. AGPx1 system RAM peaked at 266MB/s. AGPx1 fails that most basic test because the graphics card needs dedicated VRAM to keep up with the monitor.

Even more damning is that motherboards are on slowest upgrade cycle, and PC66 SDRAM of the time had a peak bandwidth of 533MB/s. Graphics cards are on the fastest upgrade cycle, and VRAM at the time was boasting 1.6GB/s bandwidth. Their different upgrade cycles means it was always economically impossible for new system RAM to catch up with new VRAM.

AGP is faster than PCI at pre-processing transactions - that's the stuff that happens before the game begins, such as loading textures. What that means is that an AGP system might start playing a game a few milliseconds sooner than a PCI system, but once you start playing.. which is once everything is VRAM.. PCI graphics cards can be identical to AGP graphics cards.

And, to maximise AGP slot stability we disable AGP chipset features: in the BIOS we select minimum aperture size, we disable sideband addressing, and that's okay because disabling AGP delivers no measurable real-world setback. If I'm honest, I wish the BIOS would allow me to set AGP aperture size to 0MB so that I can know all my system RAM to do something useful.

The only thing that materially mattered is that nobody wanted to leave their AGP slot empty, so the PCI market dried up. The real legacy of AGP is that it physically imposed exactly one graphics card slot meaning graphics APIs were forced to consolidate - why would Intel care about that? What is long forgotten is that 3Dfx were working on a motherboard chipset 😉

Well I published the photo of the ASROCK 775i65GV, it is part of a retro lot that I bought, and also includes an M209 with 286-20 and a 486-DX2-66 VLB.
I learned here, the strangeness of this ASROCK motherboard, and this similar AGP port, from what I understand, the speed is similar to an AGP video card, even using compatible AGP cards, in the manual there is a list of compatible VGA, I don't think there are flagship models, rather I imagine that they are mostly cheap models or average performance.

PC@LIVE wrote on Today, 14:50:

MattRocks wrote on Yesterday, 21:48:

There's nothing wrong with fake AGP because real AGP was a gimmick. Or, more cynically, AGP was Intel's attempt to tame the grap […]
Show full quote

PcBytes wrote on 2026-06-04, 18:18:

Purple AGP = fake AGP. That's how ASRock got AGP on non-AGP chipsets (845GV, 865GV), using their "AGI" branded slot.
https://theretroweb.com/motherboards/s/asrock … 5i65gv-rev-1-00

There's nothing wrong with fake AGP because real AGP was a gimmick. Or, more cynically, AGP was Intel's attempt to tame the graphics card industry.

Think about it. The year was 1997. Top end monitors were boasting resolutions of 1800×1440. To render basic desktop GDI at 1800x1440 you need >600MB/s. AGPx1 system RAM peaked at 266MB/s. AGPx1 fails that most basic test because the graphics card needs dedicated VRAM to keep up with the monitor.

Even more damning is that motherboards are on slowest upgrade cycle, and PC66 SDRAM of the time had a peak bandwidth of 533MB/s. Graphics cards are on the fastest upgrade cycle, and VRAM at the time was boasting 1.6GB/s bandwidth. Their different upgrade cycles means it was always economically impossible for new system RAM to catch up with new VRAM.

AGP is faster than PCI at pre-processing transactions - that's the stuff that happens before the game begins, such as loading textures. What that means is that an AGP system might start playing a game a few milliseconds sooner than a PCI system, but once you start playing.. which is once everything is VRAM.. PCI graphics cards can be identical to AGP graphics cards.

And, to maximise AGP slot stability we disable AGP chipset features: in the BIOS we select minimum aperture size, we disable sideband addressing, and that's okay because disabling AGP delivers no measurable real-world setback. If I'm honest, I wish the BIOS would allow me to set AGP aperture size to 0MB so that I can know all my system RAM to do something useful.

The only thing that materially mattered is that nobody wanted to leave their AGP slot empty, so the PCI market dried up. The real legacy of AGP is that it physically imposed exactly one graphics card slot meaning graphics APIs were forced to consolidate - why would Intel care about that? What is long forgotten is that 3Dfx were working on a motherboard chipset 😉

Well I published the photo of the ASROCK 775i65GV, it is part of a retro lot that I bought, and also includes an M209 with 286-20 and a 486-DX2-66 VLB.
I learned here, the strangeness of this ASROCK motherboard, and this similar AGP port, from what I understand, the speed is similar to an AGP video card, even using compatible AGP cards, in the manual there is a list of compatible VGA, I don't think there are flagship models, rather I imagine that they are mostly cheap models or average performance.

Are you trying to say the strange AGP-like port performs the same as an AGP slot?

I think AGI supports Radeon 9700/9800 and that would be the flagship sorted.

rasz_pl wrote on Today, 13:00:

MattRocks wrote on Today, 09:06:

The extra cache on K6-2+ means it can setup triangles multiple times faster than the K6-2.

lets get real, K6-2+ is still slower than Intels 1998 low end of a Celeron 300A. Nothing runs 'multiple times faster' on it compared to K6-2 😀 even if you gimp it by completely disabling L2 and loading motherboard with 512MB of uncached ram https://www.philscomputerlab.com/amd-k6-2-vs- … -vs-k6-iii.html
Besides two other examples use proper configs with real boy processors 😀

Please stop moving the goalposts. My original comment was on fake AGP vs real AGP, which I likened to PCI vs AGP.

Your response is on Celeron 300A vs K6-2+, and MS-DOS. Where did any of that come from? In Windows DX titles a GPU is much faster than a CPU, so the GPU is waiting on the CPU. Pure software DOS tests do not show how much time a GPU waits on the CPU to setup triangles and orchestrate Direct3D calls. And, it's all way off topic of my comment.

MattRocks wrote on Today, 17:41:

PC@LIVE wrote on Today, 14:50:

MattRocks wrote on Yesterday, 21:48:

There's nothing wrong with fake AGP because real AGP was a gimmick. Or, more cynically, AGP was Intel's attempt to tame the grap […]
Show full quote

There's nothing wrong with fake AGP because real AGP was a gimmick. Or, more cynically, AGP was Intel's attempt to tame the graphics card industry.

Think about it. The year was 1997. Top end monitors were boasting resolutions of 1800×1440. To render basic desktop GDI at 1800x1440 you need >600MB/s. AGPx1 system RAM peaked at 266MB/s. AGPx1 fails that most basic test because the graphics card needs dedicated VRAM to keep up with the monitor.

Even more damning is that motherboards are on slowest upgrade cycle, and PC66 SDRAM of the time had a peak bandwidth of 533MB/s. Graphics cards are on the fastest upgrade cycle, and VRAM at the time was boasting 1.6GB/s bandwidth. Their different upgrade cycles means it was always economically impossible for new system RAM to catch up with new VRAM.

AGP is faster than PCI at pre-processing transactions - that's the stuff that happens before the game begins, such as loading textures. What that means is that an AGP system might start playing a game a few milliseconds sooner than a PCI system, but once you start playing.. which is once everything is VRAM.. PCI graphics cards can be identical to AGP graphics cards.

And, to maximise AGP slot stability we disable AGP chipset features: in the BIOS we select minimum aperture size, we disable sideband addressing, and that's okay because disabling AGP delivers no measurable real-world setback. If I'm honest, I wish the BIOS would allow me to set AGP aperture size to 0MB so that I can know all my system RAM to do something useful.

The only thing that materially mattered is that nobody wanted to leave their AGP slot empty, so the PCI market dried up. The real legacy of AGP is that it physically imposed exactly one graphics card slot meaning graphics APIs were forced to consolidate - why would Intel care about that? What is long forgotten is that 3Dfx were working on a motherboard chipset 😉

Well I published the photo of the ASROCK 775i65GV, it is part of a retro lot that I bought, and also includes an M209 with 286-20 and a 486-DX2-66 VLB.
I learned here, the strangeness of this ASROCK motherboard, and this similar AGP port, from what I understand, the speed is similar to an AGP video card, even using compatible AGP cards, in the manual there is a list of compatible VGA, I don't think there are flagship models, rather I imagine that they are mostly cheap models or average performance.

Are you trying to say the strange AGP-like port performs the same as an AGP slot?

It doesn't.
The AGI slot on ASRock is literally AGP over PCI bus.
It even shares the same traits as PCI - 3.3v operating voltage (which is why there's a warning about not inserting 1.5v cards in that slot) and 33MHz clock instead of AGP's standard 66MHz and (in Intel 865's case) 1.5V operating voltage.

PcBytes wrote on Today, 17:58:

MattRocks wrote on Today, 17:41:

PC@LIVE wrote on Today, 14:50:

Well I published the photo of the ASROCK 775i65GV, it is part of a retro lot that I bought, and also includes an M209 with 286-20 and a 486-DX2-66 VLB.
I learned here, the strangeness of this ASROCK motherboard, and this similar AGP port, from what I understand, the speed is similar to an AGP video card, even using compatible AGP cards, in the manual there is a list of compatible VGA, I don't think there are flagship models, rather I imagine that they are mostly cheap models or average performance.

Are you trying to say the strange AGP-like port performs the same as an AGP slot?

It doesn't. The AGI slot on ASRock is literally AGP over PCI bus. It even shares the same traits as PCI - 3.3v operating voltage (which is why there's a warning about not inserting 1.5v cards in that slot) and 33MHz clock instead of AGP's standard 66MHz.

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP?

My view is that a technical comparison should boil down to 8xAGP 2133 MB/s vs PCI 133 MB/s, and to have any real-world differentiator there needs to be evidence that 133 MB/s is a bottleneck during DX9 gameplay. I have never seen any evidence of that. My challenge is to surface that evidence, which might exist! I only know that during the DX6 generation the 133 MB/s limit was more imagined bottleneck than actual bottleneck.

Radeon 9700 Pro pumps data with 19.4 GB/s VRAM bandwidth, and I suspect games running on a 9700 don't give one hoot about the bus bandwidth. And, I say that because games running on a Voodoo3 with 2.3 GB/s VRAM bandwidth didn't give one hoot about bus bandwidth.

MattRocks wrote on Today, 18:10:

PcBytes wrote on Today, 17:58:

MattRocks wrote on Today, 17:41:

Are you trying to say the strange AGP-like port performs the same as an AGP slot?

It doesn't. The AGI slot on ASRock is literally AGP over PCI bus. It even shares the same traits as PCI - 3.3v operating voltage (which is why there's a warning about not inserting 1.5v cards in that slot) and 33MHz clock instead of AGP's standard 66MHz.

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP?

Because AGI is a shared PCI device completely independent of the AGP bus altogether. While physically it's shaped as AGP, it's mapped as just another PCI slot.

On the iGP-only Intel chipsets such as 845GV and 865GV, the internal AGP bus is ALREADY attributed to the internal graphics - a key point as to why standard boards with these chipsets usually have the AGP slot section unpopulated.

ASRock's workaround involves tapping the purple AGP slot into the PCI bus, as a independent PCI device.

That's why it shares all its traits with PCI 33MHz - because outside of the slot itself, it's just a plain PCI 33MHz device, NOT an AGP device (not even AGP1x, despite 1x being the closest to PCI spec.) - hence where the "fake AGP" name from.

PcBytes wrote on Today, 18:53:

Because AGI is a shared PCI device completely independent of the AGP bus altogether. While physically it's shaped as AGP, it's m […]
Show full quote

MattRocks wrote on Today, 18:10:

PcBytes wrote on Today, 17:58:

It doesn't. The AGI slot on ASRock is literally AGP over PCI bus. It even shares the same traits as PCI - 3.3v operating voltage (which is why there's a warning about not inserting 1.5v cards in that slot) and 33MHz clock instead of AGP's standard 66MHz.

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP?

Because AGI is a shared PCI device completely independent of the AGP bus altogether. While physically it's shaped as AGP, it's mapped as just another PCI slot.

On the iGP-only Intel chipsets such as 845GV and 865GV, the internal AGP bus is ALREADY attributed to the internal graphics - a key point as to why standard boards with these chipsets usually have the AGP slot section unpopulated.

ASRock's workaround involves tapping the purple AGP slot into the PCI bus, as a independent PCI device.

That's why it shares all its traits with PCI 33MHz - because outside of the slot itself, it's just a plain PCI 33MHz device, NOT an AGP device (not even AGP1x, despite 1x being the closest to PCI spec.) - hence where the "fake AGP" name from.

So what? Where is the evidence that PCI bandwidth is a real-world bottleneck?

I am aware you could engineer a bottleneck by adding PCI SCSI SATA adapters and performing drive backups while playing a game, and I would not count that as a good faith response 😉

MattRocks wrote on Today, 18:10:

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP? […]
Show full quote

PcBytes wrote on Today, 17:58:

MattRocks wrote on Today, 17:41:

Are you trying to say the strange AGP-like port performs the same as an AGP slot?

It doesn't. The AGI slot on ASRock is literally AGP over PCI bus. It even shares the same traits as PCI - 3.3v operating voltage (which is why there's a warning about not inserting 1.5v cards in that slot) and 33MHz clock instead of AGP's standard 66MHz.

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP?

My view is that a technical comparison should boil down to 8xAGP 2133 MB/s vs PCI 133 MB/s, and to have any real-world differentiator there needs to be evidence that 133 MB/s is a bottleneck during DX9 gameplay. I have never seen any evidence of that. My challenge is to surface that evidence, which might exist! I only know that during the DX6 generation the 133 MB/s limit was more imagined bottleneck than actual bottleneck.

Radeon 9700 Pro pumps data with 19.4 GB/s VRAM bandwidth, and I suspect games running on a 9700 don't give one hoot about the bus bandwidth. And, I say that because games running on a Voodoo3 with 2.3 GB/s VRAM bandwidth didn't give one hoot about bus bandwidth.

Well I was saying that they pointed out to me here, that even if it works with some AGP video cards, the performance should be slower, but I can't say how much, and I don't know if there is such a difference between the integrated VGA, to prefer the AGP one that will still be slower.?

PC@LIVE wrote on Today, 18:58:

MattRocks wrote on Today, 18:10:

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP? […]
Show full quote

PcBytes wrote on Today, 17:58:

It doesn't. The AGI slot on ASRock is literally AGP over PCI bus. It even shares the same traits as PCI - 3.3v operating voltage (which is why there's a warning about not inserting 1.5v cards in that slot) and 33MHz clock instead of AGP's standard 66MHz.

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP?

My view is that a technical comparison should boil down to 8xAGP 2133 MB/s vs PCI 133 MB/s, and to have any real-world differentiator there needs to be evidence that 133 MB/s is a bottleneck during DX9 gameplay. I have never seen any evidence of that. My challenge is to surface that evidence, which might exist! I only know that during the DX6 generation the 133 MB/s limit was more imagined bottleneck than actual bottleneck.

Radeon 9700 Pro pumps data with 19.4 GB/s VRAM bandwidth, and I suspect games running on a 9700 don't give one hoot about the bus bandwidth. And, I say that because games running on a Voodoo3 with 2.3 GB/s VRAM bandwidth didn't give one hoot about bus bandwidth.

Well I was saying that they pointed out to me here, that even if it works with some AGP video cards, the performance should be slower, but I can't say how much, and I don't know if there is such a difference between the integrated VGA, to prefer the AGP one that will still be slower.?

To rephrase your question: Is the Radeon 9800 compatible with the AGI slot? Can a basic integrated graphics match a Radeon 9800 in DX9 gameplay?

According to the 775i65GV manual, compatible cards include CLUB3D ATI R9800, Gigabyte RADEON 9700 PRO. Either of those cards would crucify integrated graphics in DX9 gameplay. Finding a specific flavour of 9700/9800 will be frustrating though 🙁

ProLink GeForce4 Ti 4200 is listed as compatible, which is a decent DX8 card that should outpace the integrated graphics. The list even includes DX7 cards so you can reach back a fair way. I saw no DX6 AGP cards listed, but you can get genuine DX6 PCI cards so you're not actually limited by capability - you're just constrained by scarcity.

MattRocks wrote on Today, 19:01:

PC@LIVE wrote on Today, 18:58:

MattRocks wrote on Today, 18:10:

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP?

My view is that a technical comparison should boil down to 8xAGP 2133 MB/s vs PCI 133 MB/s, and to have any real-world differentiator there needs to be evidence that 133 MB/s is a bottleneck during DX9 gameplay. I have never seen any evidence of that. My challenge is to surface that evidence, which might exist! I only know that during the DX6 generation the 133 MB/s limit was more imagined bottleneck than actual bottleneck.

Radeon 9700 Pro pumps data with 19.4 GB/s VRAM bandwidth, and I suspect games running on a 9700 don't give one hoot about the bus bandwidth. And, I say that because games running on a Voodoo3 with 2.3 GB/s VRAM bandwidth didn't give one hoot about bus bandwidth.

Well I was saying that they pointed out to me here, that even if it works with some AGP video cards, the performance should be slower, but I can't say how much, and I don't know if there is such a difference between the integrated VGA, to prefer the AGP one that will still be slower.?

To rephrase your question: Is the Radeon 9800 compatible with the AGI slot? Can a basic integrated graphics match a Radeon 9800 in DX9 gameplay?

According to the 775i65GV manual, compatible cards include CLUB3D ATI R9800, Gigabyte RADEON 9700 PRO. Either of those cards would crucify integrated graphics in DX9 gameplay. Finding a specific flavour of 9700/9800 will be frustrating though 🙁

As far as it's concerned:

- 9700 Pro works since R300 uses 3.3v signaling.
- 9800s are an quite uncertain scenario. Most info I could find state R300 (9500/9700) being the last ATI card to support 3.3v AGP (which is what the "enriched PCI slot" that is AGI uses) but that some early 9800s also use it. R360 and RV350/360 are DEFINITELY NOT 3.3v compatible, I can confirm that much though.

EDIT: Ti4200 wouldn't really be a wise choice. You'll basically have it run in PCI mode which sucks bigtime.

PcBytes wrote on Today, 19:23:

As far as it's concerned: […]
Show full quote

MattRocks wrote on Today, 19:01:

PC@LIVE wrote on Today, 18:58:

Well I was saying that they pointed out to me here, that even if it works with some AGP video cards, the performance should be slower, but I can't say how much, and I don't know if there is such a difference between the integrated VGA, to prefer the AGP one that will still be slower.?

To rephrase your question: Is the Radeon 9800 compatible with the AGI slot? Can a basic integrated graphics match a Radeon 9800 in DX9 gameplay?

According to the 775i65GV manual, compatible cards include CLUB3D ATI R9800, Gigabyte RADEON 9700 PRO. Either of those cards would crucify integrated graphics in DX9 gameplay. Finding a specific flavour of 9700/9800 will be frustrating though 🙁

As far as it's concerned:

- 9700 Pro works since R300 uses 3.3v signaling.
- 9800s are an quite uncertain scenario. Most info I could find state R300 (9500/9700) being the last ATI card to support 3.3v AGP (which is what the "enriched PCI slot" that is AGI uses) but that some early 9800s also use it. R360 and RV350/360 are DEFINITELY NOT 3.3v compatible, I can confirm that much though.

EDIT: Ti4200 wouldn't really be a wise choice. You'll basically have it run in PCI mode which sucks bigtime.

They all need to run in 133MB/s PCI mode because that's what the AGI slot offers - that's not a differentiator.

And, that returns me to my original challenge: What evidence do you have that 133 MB/s PCI bus bandwidth is an actual bottleneck in real-world gaming? You've posted nothing to challenge my view that provided the game assets fit into the Ti 4200's VRAM (~8 GB/s), bus bandwidth becomes irrelevant.

MattRocks wrote on Today, 18:10:

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP? […]
Show full quote

PcBytes wrote on Today, 17:58:

MattRocks wrote on Today, 17:41:

Are you trying to say the strange AGP-like port performs the same as an AGP slot?

It doesn't. The AGI slot on ASRock is literally AGP over PCI bus. It even shares the same traits as PCI - 3.3v operating voltage (which is why there's a warning about not inserting 1.5v cards in that slot) and 33MHz clock instead of AGP's standard 66MHz.

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP?

My view is that a technical comparison should boil down to 8xAGP 2133 MB/s vs PCI 133 MB/s, and to have any real-world differentiator there needs to be evidence that 133 MB/s is a bottleneck during DX9 gameplay. I have never seen any evidence of that. My challenge is to surface that evidence, which might exist! I only know that during the DX6 generation the 133 MB/s limit was more imagined bottleneck than actual bottleneck.

Radeon 9700 Pro pumps data with 19.4 GB/s VRAM bandwidth, and I suspect games running on a 9700 don't give one hoot about the bus bandwidth. And, I say that because games running on a Voodoo3 with 2.3 GB/s VRAM bandwidth didn't give one hoot about bus bandwidth.

133MHz is a huge bottleneck for DX7+ graphics cards. Take the FX 5200 for example. I have two almost identical cards - a Zotac PCI FX 5200 (blue PCB) - 256MB of 5ns DDR over a 128 bit bus, running at 333Mhz vram and 250MHz core - and an Inno3D FX5200 "ultimate" (that's the name of the card, it's on the box) - black PCB, same exact elixir 5ns memory with the exact same layout (all 8 vram chips on the front of the card) only AGP. The inno3d card is clocked at 250mhz core and 400Mhz vram. On my Pentium 4 3400Mhz testbench (abit AS8 mainboard) the Zotac PCI card scores a whopping ~5900pts in 3D Mark 01, and that's after I overclocked the vram to 400Mhz to match the inno3d card (it's the same ram so why not give the PCI card a fair shot). The AGP Inno3D at the same clocks scored 8300 pts - 30% difference. If that's not a bottleneck, I don't know what is.

Furthermore, in UT2004 the Inno3D AGP card offers somewhat playable performance @ 1024x768 medium settings - the PCI card is unfortunately stutterfest central. Similar FPS but horrible frametimes making the game unplayable.

I'll post proof in the weekend. I have the benchmarks saved. I don't know if vogons will let me upload video, but I'll film both cards in UT2004 and try to post it. I'll make a new thread in the "video" section.

Socket3 wrote on Today, 19:53:

133MHz is a huge bottleneck for DX7+ graphics cards. Take the FX 5200 for example. I have two almost identical cards - a Zotac P […]
Show full quote

MattRocks wrote on Today, 18:10:

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP? […]
Show full quote

PcBytes wrote on Today, 17:58:

It doesn't. The AGI slot on ASRock is literally AGP over PCI bus. It even shares the same traits as PCI - 3.3v operating voltage (which is why there's a warning about not inserting 1.5v cards in that slot) and 33MHz clock instead of AGP's standard 66MHz.

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP?

My view is that a technical comparison should boil down to 8xAGP 2133 MB/s vs PCI 133 MB/s, and to have any real-world differentiator there needs to be evidence that 133 MB/s is a bottleneck during DX9 gameplay. I have never seen any evidence of that. My challenge is to surface that evidence, which might exist! I only know that during the DX6 generation the 133 MB/s limit was more imagined bottleneck than actual bottleneck.

Radeon 9700 Pro pumps data with 19.4 GB/s VRAM bandwidth, and I suspect games running on a 9700 don't give one hoot about the bus bandwidth. And, I say that because games running on a Voodoo3 with 2.3 GB/s VRAM bandwidth didn't give one hoot about bus bandwidth.

133MHz is a huge bottleneck for DX7+ graphics cards. Take the FX 5200 for example. I have two almost identical cards - a Zotac PCI FX 5200 (blue PCB) - 256MB of 5ns DDR over a 128 bit bus, running at 333Mhz vram and 250MHz core - and an Inno3D FX5200 "ultimate" (that's the name of the card, it's on the box) - black PCB, same exact elixir 5ns memory with the exact same layout (all 8 vram chips on the front of the card) only AGP. The inno3d card is clocked at 250mhz core and 400Mhz vram. On my Pentium 4 3400Mhz testbench (abit AS8 mainboard) the Zotac PCI card scores a whopping ~5900pts in 3D Mark 01, and that's after I overclocked the vram to 400Mhz to match the inno3d card (it's the same ram so why not give the PCI card a fair shot). The AGP Inno3D at the same clocks scored 8300 pts - 30% difference. If that's not a bottleneck, I don't know what is.

Furthermore, in UT2004 the Inno3D AGP card offers somewhat playable performance @ 1024x768 medium settings - the PCI card is unfortunately stutterfest central. Similar FPS but horrible frametimes making the game unplayable.

I'll post proof in the weekend. I have the benchmarks saved. I don't know if vogons will let me upload video, but I'll film both cards in UT2004 and try to post it. I'll make a new thread in the "video" section.

" Zotac PCI card scores a whopping ~5900pts in 3D Mark 01 ... AGP Inno3D at the same clocks scored 8300 pts"

In other words, the slower PCI bus bandwidth is not the limiting factor? You appear to be strengthening my argument by saying a PCI card outpaces the AGP card - exactly the same thing we saw with Voodoo3 PCI cards outpacing Voodoo3 AGP cards.

I'm confused with this 3.3 volt talk, the board PC@LIVE posted the manual states "ASRock Graphics Interface Slot (1.5V_AGP1) "

Grem Five wrote on Today, 20:04:

I'm confused with this 3.3 volt talk, the board PC@LIVE posted the manual states "ASRock Graphics Interface Slot (1.5V_AGP1) "

I don't know. I steered away from voltages - just saying the plastic slot itself should physically prevent users inserting cards that are electrically incompatible. I'm trying to focus on PCI/AGP bandwidth, when it matters, and when it doesn't.

MattRocks wrote on Today, 20:03:

Socket3 wrote on Today, 19:53:

133MHz is a huge bottleneck for DX7+ graphics cards. Take the FX 5200 for example. I have two almost identical cards - a Zotac P […]
Show full quote

MattRocks wrote on Today, 18:10:

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP?

My view is that a technical comparison should boil down to 8xAGP 2133 MB/s vs PCI 133 MB/s, and to have any real-world differentiator there needs to be evidence that 133 MB/s is a bottleneck during DX9 gameplay. I have never seen any evidence of that. My challenge is to surface that evidence, which might exist! I only know that during the DX6 generation the 133 MB/s limit was more imagined bottleneck than actual bottleneck.

Radeon 9700 Pro pumps data with 19.4 GB/s VRAM bandwidth, and I suspect games running on a 9700 don't give one hoot about the bus bandwidth. And, I say that because games running on a Voodoo3 with 2.3 GB/s VRAM bandwidth didn't give one hoot about bus bandwidth.

133MHz is a huge bottleneck for DX7+ graphics cards. Take the FX 5200 for example. I have two almost identical cards - a Zotac PCI FX 5200 (blue PCB) - 256MB of 5ns DDR over a 128 bit bus, running at 333Mhz vram and 250MHz core - and an Inno3D FX5200 "ultimate" (that's the name of the card, it's on the box) - black PCB, same exact elixir 5ns memory with the exact same layout (all 8 vram chips on the front of the card) only AGP. The inno3d card is clocked at 250mhz core and 400Mhz vram. On my Pentium 4 3400Mhz testbench (abit AS8 mainboard) the Zotac PCI card scores a whopping ~5900pts in 3D Mark 01, and that's after I overclocked the vram to 400Mhz to match the inno3d card (it's the same ram so why not give the PCI card a fair shot). The AGP Inno3D at the same clocks scored 8300 pts - 30% difference. If that's not a bottleneck, I don't know what is.

Furthermore, in UT2004 the Inno3D AGP card offers somewhat playable performance @ 1024x768 medium settings - the PCI card is unfortunately stutterfest central. Similar FPS but horrible frametimes making the game unplayable.

I'll post proof in the weekend. I have the benchmarks saved. I don't know if vogons will let me upload video, but I'll film both cards in UT2004 and try to post it. I'll make a new thread in the "video" section.

" Zotac PCI card scores a whopping ~5900pts in 3D Mark 01 ... AGP Inno3D at the same clocks scored 8300 pts"

In other words, the slower PCI bus bandwidth is not the limiting factor? You appear to be strengthening my argument by saying a PCI card outpaces the AGP card - exactly the same thing we saw with Voodoo3 PCI cards outpacing Voodoo3 AGP cards.

PCI 5900 points
AGP 8300 points

Exactly not what you said 😉

MattRocks wrote on Today, 20:03:

Socket3 wrote on Today, 19:53:

133MHz is a huge bottleneck for DX7+ graphics cards. Take the FX 5200 for example. I have two almost identical cards - a Zotac P […]
Show full quote

MattRocks wrote on Today, 18:10:

Ok, but AGP 1.0 was 3.3V too. Is there any objection to treating AGI vs AGP as PCI vs AGP?

My view is that a technical comparison should boil down to 8xAGP 2133 MB/s vs PCI 133 MB/s, and to have any real-world differentiator there needs to be evidence that 133 MB/s is a bottleneck during DX9 gameplay. I have never seen any evidence of that. My challenge is to surface that evidence, which might exist! I only know that during the DX6 generation the 133 MB/s limit was more imagined bottleneck than actual bottleneck.

Radeon 9700 Pro pumps data with 19.4 GB/s VRAM bandwidth, and I suspect games running on a 9700 don't give one hoot about the bus bandwidth. And, I say that because games running on a Voodoo3 with 2.3 GB/s VRAM bandwidth didn't give one hoot about bus bandwidth.

133MHz is a huge bottleneck for DX7+ graphics cards. Take the FX 5200 for example. I have two almost identical cards - a Zotac PCI FX 5200 (blue PCB) - 256MB of 5ns DDR over a 128 bit bus, running at 333Mhz vram and 250MHz core - and an Inno3D FX5200 "ultimate" (that's the name of the card, it's on the box) - black PCB, same exact elixir 5ns memory with the exact same layout (all 8 vram chips on the front of the card) only AGP. The inno3d card is clocked at 250mhz core and 400Mhz vram. On my Pentium 4 3400Mhz testbench (abit AS8 mainboard) the Zotac PCI card scores a whopping ~5900pts in 3D Mark 01, and that's after I overclocked the vram to 400Mhz to match the inno3d card (it's the same ram so why not give the PCI card a fair shot). The AGP Inno3D at the same clocks scored 8300 pts - 30% difference. If that's not a bottleneck, I don't know what is.

Furthermore, in UT2004 the Inno3D AGP card offers somewhat playable performance @ 1024x768 medium settings - the PCI card is unfortunately stutterfest central. Similar FPS but horrible frametimes making the game unplayable.

I'll post proof in the weekend. I have the benchmarks saved. I don't know if vogons will let me upload video, but I'll film both cards in UT2004 and try to post it. I'll make a new thread in the "video" section.

" Zotac PCI card scores a whopping ~5900pts in 3D Mark 01 ... AGP Inno3D at the same clocks scored 8300 pts"

In other words, the slower PCI bus bandwidth is not the limiting factor? You appear to be strengthening my argument by saying a PCI card outpaces the AGP card - exactly the same thing we saw with Voodoo3 PCI cards outpacing Voodoo3 AGP cards.

Is this a joke or something? How do you see 5900 pts as being higher than 8300 pts? He clearly said which one was PCI. The AGP card is 40% faster at the same clocks.

Oh, dyslexic. I saw 5900 as 9500! 🤣

Ok, I look forward to seeing the benchmark results and motherboard settings. But, if it's stuttering then that suggests to me some assets are not in local VRAM? I imagine the FX5200 PCI has some extra silicone near the PCI fingers too? It's an interesting example though. Would be good to see older 3DMark as well.