shevalier wrote on 2023-10-18, 07:45:

Dual channel makes sense for any parallel process that accesses memory past the CPU, for example loading textures into video mem […]
Show full quote

CharlieFoxtrot wrote on 2023-10-18, 04:21:

That dual channel memory controller is completely useless on these, unless you use the GF4MX integrated graphics available on some boards. The whole idea of the solution was to reserve the other channel for IGP.

Dual channel makes sense for any parallel process that accesses memory past the CPU, for example loading textures into video memory during intensive exchange with the memory of the CPU itself.
Or copying from the network to disk under similar conditions.
The difference is not +100, but a modest +10-15.
Plus prefetch in the northbridge (sometimes called L3 cache). Another 10 percent productivity.
At its peak, the difference between nF2 and KT400/600 reached 30 percent. Sometimes even 0, but on average - around 15 percent.
But in those days this was the difference (relatively speaking) between Athlone 2200+ and 2500+.
Which also cost money.

Most of the nForce2 edge over KT400/600 comes from the other factors of the chipset, not from the ability to use dual channel, unless you use IGP.

Here is an old test about the nForce dual channel vs KT400 and performance difference for most tests is low single digits. Where the difference is bigger, it is explained in the article. I'd say 15% is quite a stretch in real world scenarios, especially gaming.

http://ixbtlabs.com/articles2/nforce2-1vs2cha … nels/index.html

The problem with Athlon XP was that @200 is FSB output 3,2GB/s while dual channel 1:1 provides double the memory bandwidth. That is also the reason why IGP gets by far the biggest benefit from it as CPU can't saturate the 6,4GB/s maximum effectively that is available to it.

dionb wrote on 2023-10-18, 08:49:

An anecdote: back in the day, my flatmate (who had rich parents and didn't spend as much on travel and drinks) bought a shiny new A7N8X-Deluxe,

This is not a anecdote, this is Asus as is.
Motherboard based on nForce 3 and A64 s754.
Memory from QVL.
But there is no boot from USB flash drive. 2006 (?), but there is no boot from USB.
BIOS update is coming out.
A boot from USB is appears, and the memory starts to fill with errors.

CharlieFoxtrot wrote on 2023-10-18, 08:57:

Here is an old test about the nForce dual channel vs KT400 and performance difference for most tests is low single digits.

This is semi-synthetic testing.
In real operation, there were always background processes that loaded the system in parallel.
The same antivirus caught on a file.
Or network activity. Or less micro freeze when loading textures in games. In bare numbers this is not measurable.
Anything that used DMA suffered a smaller penalty when paralleling high EV6 CPU-chipset bus load.

Dual channel is not a silver bullet, but a very nice addition to the functionality of the chipset.

I tend to avoid A7N8X/-E Deluxe boards. ASUS couldn't be arsed to add a 12v plug and the whole board feels awfully cheap (like below ECS level of crap).
One wrong move and the southbridge goes dead. (I have had a -E Deluxe southbridge die out of the blue).

Epox 8RDA6+ Pro is my go to board for the nF2 platform. NF7 would have been ideal but I had zero luck finding one.

PcBytes wrote on 2023-10-18, 10:25:

I tend to avoid A7N8X/-E Deluxe boards. ASUS couldn't be arsed to add a 12v plug and the whole board feels awfully cheap (like below ECS level of crap).
One wrong move and the southbridge goes dead. (I have had a -E Deluxe southbridge die out of the blue).

Epox 8RDA6+ Pro is my go to board for the nF2 platform. NF7 would have been ideal but I had zero luck finding one.

Considering that none of the Asus fans reacted to the phrase Bus disconnect, it is implied that if Asus did not implement the feature, then it does not exist.
http://www.amd.com/us-en/assets/content_type/ … _docs/26237.pdf
p 21 "Power Management"
And then urban legends appear that AXP is a very hot processor.
Although it temperature at rest is the same as that of the surrounding air. 😀

shevalier wrote on 2023-10-18, 10:07:

This is not a anecdote, this is Asus as is. Motherboard based on nForce 3 and A64 s754. Memory from QVL. But there is no boot f […]
Show full quote

dionb wrote on 2023-10-18, 08:49:

An anecdote: back in the day, my flatmate (who had rich parents and didn't spend as much on travel and drinks) bought a shiny new A7N8X-Deluxe,

This is not a anecdote, this is Asus as is.
Motherboard based on nForce 3 and A64 s754.
Memory from QVL.
But there is no boot from USB flash drive. 2006 (?), but there is no boot from USB.
BIOS update is coming out.
A boot from USB is appears, and the memory starts to fill with errors.

CharlieFoxtrot wrote on 2023-10-18, 08:57:

Here is an old test about the nForce dual channel vs KT400 and performance difference for most tests is low single digits.

This is semi-synthetic testing.
In real operation, there were always background processes that loaded the system in parallel.
The same antivirus caught on a file.
Or network activity. Or less micro freeze when loading textures in games. In bare numbers this is not measurable.
Anything that used DMA suffered a smaller penalty when paralleling high EV6 CPU-chipset bus load.

Dual channel is not a silver bullet, but a very nice addition to the functionality of the chipset.

It is semi synthetic, but it is there show the difference and effect of the dual channel on the platform in clear comparable raw numbers and in different test scenarios. It is better than "I feel that this is snappier" and it allows the evaluation of the behavior of dual channel in different situations.

I haven't clamed that it is not a nice addition, but nVidias aim with dual channel was indeed to improve the performance of IGP, rest is just extra which is not much with Athlon XP. There are countless of old forum discussions and tests dealing with this still readable and consensus is that the difference is insignificant. If you wanted to max your performance with overclocking, it was much better option to go with one single high-end memory stick (ie single channel mode), which helped overclocking the FSB and memory 1:1 as high as possible, as with dual stick configuration you usually couldn't go that high and/or needed to settle with looser timings. Higher FSB/memory frequency quite easily beated the benefit from dual channel.

And indeed, it wasn't always an improvement either. I just searched this article, because I remembered there was some buzz back in the day about the performance of cheaper nForce2 400 compared to full fledged nForce 400 Ultra. NV400 has only single channel memory controller, while the top of the line Ultra supported dual channel operation. Still, in this quite comprehensive test the cheapo Soltek with NV400 beated the popular Asus A7N8X in most of the tests, sometimes even with quite clear difference:
https://web.archive.org/web/20031015165720/ht … ID=244&PageID=1

Ultra wasn't slow because it used dual channel, but more so, because the humble NV400 memory controller was indeed better and probably has lower latencies than the more complex dual channel controller on the Ultra.

One of the best computers I ever had was a Barton Athlon XP-M conservatively overclocked at 3000+ speed on an ABit AN7. Everything worked perfectly on that board, it was just a dream to use. Functionally it was better than my modern PC.
I also tried an NF7-S v2.0 with the same parts. That board came at the right time to be a huge hit with overclockers on the internet. The AN7 was a late board that replaced it, but didn't wow the overclockers who already had NF7-S boards, so that group kept promoting the older board.

With my pair of boards, I didn't see any advantage to the NF7-S. For me they were equal in overclocking, but the AN7 had some more modern features that I liked.
The AN7 has temperature based fan control that the NF7-S lacks. The AN7 can boot from USB, I think the NF7-S cannot.
The AN7 can save a few different overclocking "profiles" in the BIOS, so you can load different settings to run at a different speed. Maybe not that useful, but could be if you have some reason to want to slow the system down.
It's been a long time so I might be forgetting some differences. It's nothing huge, but the AN7 is just a more refined board IMO.

If I was shopping today I'd take either of them depending on price.

shamino wrote on 2023-10-18, 11:28:

One of the best computers I ever had was a Barton Athlon XP-M conservatively overclocked at 3000+ speed on an ABit AN7. Everyth […]
Show full quote

One of the best computers I ever had was a Barton Athlon XP-M conservatively overclocked at 3000+ speed on an ABit AN7. Everything worked perfectly on that board, it was just a dream to use. Functionally it was better than my modern PC.
I also tried an NF7-S v2.0 with the same parts. That board came at the right time to be a huge hit with overclockers on the internet. The AN7 was a late board that replaced it, but didn't wow the overclockers who already had NF7-S boards, so that group kept promoting the older board.

With my pair of boards, I didn't see any advantage to the NF7-S. For me they were equal in overclocking, but the AN7 had some more modern features that I liked.
The AN7 has temperature based fan control that the NF7-S lacks. The AN7 can boot from USB, I think the NF7-S cannot.
The AN7 can save a few different overclocking "profiles" in the BIOS, so you can load different settings to run at a different speed. Maybe not that useful, but could be if you have some reason to want to slow the system down.
It's been a long time so I might be forgetting some differences. It's nothing huge, but the AN7 is just a more refined board IMO.

If I was shopping today I'd take either of them depending on price.

I had AN7 back in the day and I got it immediately after it was relesed and had high hopes for it. At least my sample was a complete turd and I couldn't get it stable even stock with more than one RAM sticks. It was incredibly unstable board, but on paper it was really good. I RMAed the board and got NF7-S V2.0 instead, which was my last sA board (until nowadays).

I wasn't the only one who experienced this as when I returned it to the shop after struggling it with few days, seller didn't squirm at all with the problems and said that other's have complained about the board too. Maybe it was just a buggy BIOS, but it was still so disappointing product because Abit had such a good reputation. It was just released so there was no newer versions available, so there's that. I ran NF7-S v2.0 with 230MHz+ FSB completely stable. I don't exactly remember the figure anymore, but I remember that I couldn't get it stable over 240MHz with my memory sticks.

AN7 is indeed quite forgotten MB nowadays and based on how many those are for sale nowadays, I don't think it sold that well. It may be because it wasn't well received or because it came out too late.

Based on my own experience, I don't recommend AN7, but that may be just early BIOS issue.

Edit: just googled old review of AN7. The experience was the same than mine, it was unstable and very picky about the memory used and from the three tested RAM sets, only Corsair PC3700 worked well and from the overclocking perspective it was said to be very average board at best. However, CPU clocked really well because of the good voltage controls.

You just summarized my experiences with at least three A7N8X boards. One was the cut down A7N8X-X version so I can't really complain, but the other two were the Deluxe variants.

Both would give me never-ending headaches under XP (one of them did run 7 at one point), and it was awfully picky to what RAM it liked - dunno who said the QVL sticks are guaranteed to work on it because NOT EVEN THOSE WORKED.
The capacitors weren't great either on those - the cut down A7N8X came with blown Nichicons, the Deluxes both came with blown Chemicon KZG. All three had their caps replaced with Panasonic FL caps.

I sold one of the Deluxes, the other one died, and the cut down A7N8X is in storage with some subpar Canicon caps I installed on it before sending it to storage. I'm glad to have replaced BOTH Deluxe boards with an Epox EP-8RDA6+ Pro and a MSI K7N2 Delta-ILSR. While the MSI tends to be ever so slightly picky about RAM sticks (although not so much as the ASUS), the Epox will eat absolutely anything you throw at it. The only differences I can pinpoint between the K7N2 and the 8RDA6 is the latter has native SATA support, besides the Sil3114 controller (which IMO is better than the 3112 on the ASUS), and the 8RDA6 has 7.1 channel audio, whereas the MSI has 5.1.

CharlieFoxtrot wrote on 2023-10-18, 12:02:

I had AN7 back in the day and I got it immediately after it was relesed and had high hopes for it. At least my sample was a comp […]
Show full quote

shamino wrote on 2023-10-18, 11:28:

One of the best computers I ever had was a Barton Athlon XP-M conservatively overclocked at 3000+ speed on an ABit AN7. Everyth […]
Show full quote

One of the best computers I ever had was a Barton Athlon XP-M conservatively overclocked at 3000+ speed on an ABit AN7. Everything worked perfectly on that board, it was just a dream to use. Functionally it was better than my modern PC.
I also tried an NF7-S v2.0 with the same parts. That board came at the right time to be a huge hit with overclockers on the internet. The AN7 was a late board that replaced it, but didn't wow the overclockers who already had NF7-S boards, so that group kept promoting the older board.

With my pair of boards, I didn't see any advantage to the NF7-S. For me they were equal in overclocking, but the AN7 had some more modern features that I liked.
The AN7 has temperature based fan control that the NF7-S lacks. The AN7 can boot from USB, I think the NF7-S cannot.
The AN7 can save a few different overclocking "profiles" in the BIOS, so you can load different settings to run at a different speed. Maybe not that useful, but could be if you have some reason to want to slow the system down.
It's been a long time so I might be forgetting some differences. It's nothing huge, but the AN7 is just a more refined board IMO.

If I was shopping today I'd take either of them depending on price.

I had AN7 back in the day and I got it immediately after it was relesed and had high hopes for it. At least my sample was a complete turd and I couldn't get it stable even stock with more than one RAM sticks. It was incredibly unstable board, but on paper it was really good. I RMAed the board and got NF7-S V2.0 instead, which was my last sA board (until nowadays).

I wasn't the only one who experienced this as when I returned it to the shop after struggling it with few days, seller didn't squirm at all with the problems and said that other's have complained about the board too. Maybe it was just a buggy BIOS, but it was still so disappointing product because Abit had such a good reputation. It was just released so there was no newer versions available, so there's that. I ran NF7-S v2.0 with 230MHz+ FSB completely stable. I don't exactly remember the figure anymore, but I remember that I couldn't get it stable over 240MHz with my memory sticks.

AN7 is indeed quite forgotten MB nowadays and based on how many those are for sale nowadays, I don't think it sold that well. It may be because it wasn't well received or because it came out too late.

Based on my own experience, I don't recommend AN7, but that may be just early BIOS issue.

Edit: just googled old review of AN7. The experience was the same than mine, it was unstable and very picky about the memory used and from the three tested RAM sets, only Corsair PC3700 worked well and from the overclocking perspective it was said to be very average board at best. However, CPU clocked really well because of the good voltage controls.

That's odd.. but I've only ever had the 1 board. It came complete with the box, unless I'm mixing it up in my memory I think it was an essentially NOS "store return" board. I was scrounging from mismatched parts, so I think I started with some average DDR266 with an older CPU, but long term I used the Barton Mobile with a pair of Kingston HyperX DDR400 CL2. If RAM issues were common, maybe early BIOSes didn't respond properly to some SPD values or something? I dunno.
I was squeamish about voltages so I didn't overclock real high, but within the range I was comfortable, I got the same results with both boards. On both boards I could pass tests at 3200+ speeds, but not with +5% added to the FSB (a requirement for me to consider it rock solid). So I settled at 3000+ settings.
I didn't raise the FSB much - only to 210 in stability testing (for the 5% margin test), then when that passed I set it back to 200. My CPU was unlocked so I didn't need to overclock the FSB, and preferred not to.

I do remember reading in forums a widespread preference for the NF7-S v2.0 over the AN7, but it kind of struck me as groupthink and I was skeptical of there really being much difference. But I've only had the 1 board, maybe if I had more examples, or was more aggressive with overclocking, I could have started to see it differently.
I also had better luck with a few examples of the Asus P2B-F over the P3B-F boards I've had, and most people would put it the other way around. So luck of the draw is a big element I guess.

I wish I still had the NF7-S v2.0. I gave that board to my nephew for a budget gaming build back then. Being a kid who wanted to play newer games, I pushed his overclock a little too hard trying to help it keep up, and had to dial it down later.