I can trust you guys to answer this question correctly...

[K.I.L.E.R]

about Fastwrites.

I keep hearing things like:
It slows down some games. (god knows how?)
It's unstable. (I understand some systems are unstable with it on)

That's about it.

I have no instability nor slow games running FW, FW strap and SBA all enabled at the same time.

What I would like to know how FW could slow down a system if it has proper hardware support?

Fast Writes improves all writes from the CPU to the graphics chip.
Fast Writes enables the CPU to send data directly to the graphics bus without going through the system memory. This benefits overall system performance in two ways:

*Removes the bottleneck previously that exists in systems without Fast Writes.

* Frees up system memory bus to perform other functions and increase system efficiency.

To try and answer my question "What I would like to know how FW could slow down a system if it has proper hardware support?", well the answer is that it can't. If the hardware has full and proper hardware support then it can't slow down with FW enabled.
When an Ati representative (sireric I think) stated that the reason stuttering was ocuring was because of how many tasks are polling the interrupts? (correct me if I am wrong). Since I have FW enabled, that could mean that my graphics subsytem would not have stalled and hence I never had any stuttering. All because the crucial data was directly sent from CPU to video card avoiding the usual data path.

Can anyone please tell me if my explanation is wrong?

 

[Dio]

I think it's rather more complicated than that.

Most importantly, fastwrites are out on the edges of the AGP spec - and historically, anything on the edge of the spec has always been... err... temperamental (in the very early days it was execute-mode texturing, then sideband, then 4X, then fastwrites...) and so very dependent on the combination of chipset, motherboard, video card, CPU, phase of the moon, wind speed, etc....

The thing about polling is also complex. A PC has a huge number of 'asynchronous' systems - CPU, northbridge, DRAM, AGP, PCI, southbridge, etc. Each of these has its own 'rights' to resources. Fastwrites still have to go through the northbridge (and probably the CPU write-combining buffers, which add another semi-asynchronous layer because their flushes are difficult to predict) and don't get special priority. And the video card may be in the middle of a fetch or even a queued sequence of fetches from DRAM, using up the AGP bus. Then, the CPU gets interrupted by a PCI device... you see, it's all a huge tangle that has to be sorted out.

Personally I'm astonished the PC manages to turn on in the mornings let alone get on with that lot.

 

[darkblu]

Personally I'm astonished the PC manages to turn on in the mornings let alone get on with that lot.

above qualifies for the 'thought of the day' award

 

[PSarge]

Re: I can trust you guys to answer this question correctly..

When an Ati representative (sireric I think) stated that the reason stuttering was ocuring was because of how many tasks are polling the interrupts? (correct me if I am wrong). Since I have FW enabled, that could mean that my graphics subsytem would not have stalled and hence I never had any stuttering. All because the crucial data was directly sent from CPU to video card avoiding the usual data path.

If the CPU is having to Poll the graphics card, then it's having to read repeatedly over the AGP bus. Unfortunatley fast writes are called fast writes for a reason. Basically the AGP bus goes into a burst mode when lots of writes from CPU to [GV]PU occur. Putting a read into the middle of it breaks up the burst. Putting lots of spaced out reads (i.e. reads with gaps in between, not reads that have been smoking something ) kills lots of bursts and so can kill the transfer rate.

This may be what he was talking about.

 

[Dio]

Personally I'm astonished the PC manages to turn on in the mornings let alone get on with that lot.

above qualifies for the 'thought of the day' award

Well, as discussed here some moons ago, my mind focuses on this most days given that I have a PC that frequently doesn't turn on in the mornings without careful coaxing.

My second PC that didn't turn on most mornings (or in fact most afternoons either especially if it had 1066 RDRAM in it), has been fixed by a change of motherboard (for an identical one, so next job is to talk them into taking the first one back after 8 months). Now I just have to work out why it will only find the network card if there is a floppy drive installed....

 

[Joe DeFuria]

Personally I'm astonished the PC manages to turn on in the mornings let alone get on with that lot.

Forgive me....OT....

Given that astonishment of turning on a PC, think about all the systems...electrical, chemical, and mechanical, on something like the Space Shuttle. And think abnout the conditions they are subject to.

And some people wonder "how tragedies like the Challenger and Columbia can happen." It doesn't matter how much money and resources you throw at at, all you can do is hope to bring the risk down to a reasonable level.

 

[RussSchultz]

I think they had their losses early on behind the iron curtain.

 

[darkblu]

Well, as discussed here some moons ago, my mind focuses on this most days given that I have a PC that frequently doesn't turn on in the mornings without careful coaxing.

My second PC that didn't turn on most mornings (or in fact most afternoons either especially if it had 1066 RDRAM in it), has been fixed by a change of motherboard (for an identical one, so next job is to talk them into taking the first one back after 8 months). Now I just have to work out why it will only find the network card if there is a floppy drive installed....

i was only agreeing to what you said. we tend to take all those "subtle, suitable regularities" in our lives for granted, and only few of us stop to realize (unless misfortune strikes) that all those "little" things happening to work happily for us is much more a matter of statistical luck (or god's mercy, if you prefer) than a regularity (regardless of the thing being a product of many people's hard work or not), thus my raction to your post. i, personally, am astonished to see me waking up yet another morning, given how many things may happen that could bring to the failure of my morning POST (and no, i'm not of worse health/phisical shape than the average male at my age)

 

[Tagrineth]

About Fast Writes, it looks like SMARTGART disabled them on my system. At first they were enabled of course, but...

Well anyway, I also at first didn't have OpenGL stutter, but now I do. I don't know if there's any correlation... I'm going to try re-enabling FW and running some OpenGL games, probably this weekend or something.

 

[Dio]

I think they had their losses early on behind the iron curtain.

Sorry for losing the post you were replying to. I decided I was perhaps being insensitive

 

[Randell]

yep (sorry for satying OT, but I thought I'd add)- in perspective USA has a much better safety record. Russian space travel has a history of massive acccidents. 2 launch pad explosions killed 90 (including the head of the space agency) then 50 engineering and technical staff respectively. Another time all the soyuz crew asphyxiated on reentry.

 

[flf]

My wife's system froze freqently. I yanked the old 300Watt PSU and installed and Enermax Whipser 465Watt PSU. Stable as a frickin' rock ever since. I believe that a poor quality power supply can amplify the quirks and timing problems inherent in all systems.

With regards to the shuttle...

A Remington 7MM Magnum bullet has a typical muzzel velocity of 3200 feet per second.

The "speed of sound" (as often quoted) is 968 feet per second at sea level in "average" conditions.

The Shuttle was travelling at Mach 16 or so, which is about 15488 feet per second. (Well, it was slowing down... it started out around Mach 20-something before entry.)

Put simply, the Shuttle is going really, really damn fast. I'm sure our ancestors will look back at how we do space flight and say, "Can you believe they did this crap? I mean... they just cruised straight in like a frickin' meteor!"

When you play with forces this great it's always a crapshoot.

 

[K.I.L.E.R]

Thanks to all the people who tried to help me, as this thread seems to be going off topic. I am requesting that a moderator close it.
Thank You.

 

[Dio]

My wife's system froze freqently. I yanked the old 300Watt PSU and installed and Enermax Whipser 465Watt PSU. Stable as a frickin' rock ever since. I believe that a poor quality power supply can amplify the quirks and timing problems inherent in all systems.

The aforementioned boot-problem machine has a 465W Enermax in too

Rock solid stable when I can get the damn thing to boot. In fact, I don't think it's crashed on me in three years except the two times when the motherboard has eaten the RAM. Since I replaced it with cheaper RAM last time it's stopped happening. Go figure.

Yes, long, long way OT.

 

[Tahir2]

Dio WRT 'them' taking your old motherboard back.. if they can verify the fault then they will take it back and offer you a replacement board or you may decide to take credit and buy something more expensive from them (they hate it when customers ask for refunds and I know places like ebuyer do not give refunds on items over 30 days old.. just replacement units).

Anyway I hope that info helps.. not like you actually asked or showed any inclination to requiring any help in this matter - but I was bored and feeling left out.