preferred refresh rate for gaming?

DriveEuro · Mar 26, 2003, 04:22 PM

Quote:

Originally posted by BWX232
Drive is a rich kid who like to make fun of people's old harware! LoL

Nah, this stuff is from money that I make. The only thing in here I didn't pay for is one of my Western Digital drives. So nah!

~~shai~~ · Mar 26, 2003, 05:49 PM

well your a rich kid who pays for his own stuff...

-={420}=- · Mar 27, 2003, 03:22 AM

Quote:

Originally posted by DriveEuro
Packard bell? HAHAHHAHA man, that thing has to be aged.

It's from 1998 but it's been working just fine. I do plan to get a better one as soon as I get some money.

BWX · Mar 27, 2003, 06:09 AM

Quote:

Originally posted by DriveEuro
Nah, this stuff is from money that I make. The only thing in here I didn't pay for is one of my Western Digital drives. So nah!

I was just kidding- really-

Vampyromaniac · Mar 27, 2003, 08:08 AM

Quote:

Originally posted by Matth
As far as I know, the frame buffer IS part of the video card's RAM pool

At 1024x768x32 bit (eg 3Dmark 2001), ONE frame buffer is 3 Megabytes - since it will be double bufferring, that means that two lots of 3Mb are used.
That 3Mb is pulled from the card's video RAM (and since dual ported ram basically gave way to conventional SDRAM or DDR, over tha same memory bus that is used for processing on the card) and sent at the refreash rate.

So at a low 60Hz, it's using 180Mb/s to keep the display going
At 85 Hz, it rises to 255Mb/s
and if you use 1280x1024 at 85Hz, it's 425Mb/s

Sure, it won't bother the more top-flight cards much, but you'd sure as hell notice it on a weakling like A Geforce 2MX200. If the difference between the highest allowable, and the lowest usable refresh rate is an appreciable proportion of the card's memory bandwith, you should find a difference comparable to the same degree of memory overclocking.

Ah, I see
I thought it would be separate

But, if the frame hasn't changed since the last read, does it still transfer the image?
Like, say the refresh rate is 120Hz, but the game is running at 60fps . . . will it transfer 60 times, or 120?
If it's only 60, then setting the refresh rate higher than 60 (in that example) wouldn't hurt anything, right?

Even if not . . . let me do some math . . . Say, 1280x1024x32 @ 85 = 425MB/s, and @100Hz it is: 500MB/s
The difference is only 75 MB/s ! Even on my weak card, that's only 1.4% of my bandwidth I would be saving.
And even if that corresponds to a 1.4% gain in framerate (it is probably less,) a game running at 75 fps would only gain 1 fps

I think it doesn't really matter, run at whatever refresh rate you want, there's no significant difference.
You won't notice more fps by lowering the refresh rate, but if you can make the image look more solid by raising it, do it!

Matth · Mar 28, 2003, 09:23 PM

Just ran 2 tests on my Geforce 2MX (a Creative DDR version, at stock speeds 175/143 **YES, they did use slow DDR, and it does perform worse than a good SDR model)

At 75Hz, it scored 2248, while at 60Hz, it scored 2303 (3Dmark2001SE)
A 2.4% difference, for actually somewhat less than that in bandwith terms.
- I forgot - the bandwith saved need to be taken as a percentage of the bandwith LEFT after subtracting screen refresh usage from the total - even allowing for that, it's still more than 1:1 - presumably the overhead on the GPU is also reduced.

Sure, I can gain the same and more by a RAM overclock - in fact, the score rises pretty well pro rata, so even though this is not the memory bandwidth crippled MX200, it's still memory bandwidth limited.

If you refresh at 120Hz, while getting 60 FPS, the frame buffer is being read from the card RAM 120 times per second, and that is RAM bandwith not available for any other use.

PS. Those scan rates are the limit of my LCD monitor, so while 60Hz looks totally solid (though vsync disabled tearing is actually far more objectionable than at 75), I can't go higher or lower to stretch the point.

I'll lay the challenge - with a better card, and a monitor that can do it....
1. Your normal refresh rate
2. The highest acceptable for your Monitor/card
3. 87i (Interlaced), equal to 43 full frames per second (or 60Hz if you can't)

And if you CAN use 87i, and you gain any increase, is there anything in the results to show it - or is it a "not quite cheating" way to grab another fraction of a percent an climb the results table?

Vampyromaniac · Mar 29, 2003, 02:19 AM

Quote:

Originally posted by Matth
Just ran 2 tests on my Geforce 2MX (a Creative DDR version, at stock speeds 175/143 **YES, they did use slow DDR, and it does perform worse than a good SDR model)

OK, for that card and that test, you gained 4.9% memory bandwidth, and 3dmark rose 2.4%.
Still, I'm saying even 2.4% is not something you're gonna notice in the real world, and if you notice 75Hz looks better than 60Hz you should use 75.
But if you mean for raising your benchmark scores sure why not, but shuki was asking about real life gaming

Of course with an LCD, it's different (dunno for sure how THAT goes) . . . I was always thinking of CRT's

Quote:

If you refresh at 120Hz, while getting 60 FPS, the frame buffer is being read from the card RAM 120 times per second, and that is RAM bandwith not available for any other use.

Thanks for that info

Quote:

I'll lay the challenge - with a better card, and a monitor that can do it....
1. Your normal refresh rate
2. The highest acceptable for your Monitor/card
3. 87i (Interlaced), equal to 43 full frames per second (or 60Hz if you can't)

I'm sorry if I don't understand, but I'll answer what I think you want . . .

1. 120Hz
2. if you mean highest refresh rate I can get with my monitor&vidcard combination by lowering resolution, it is: 160Hz
3. if you are asking if I have 87i available, no I don't have any interlaced available

EDIT: Ahh, reading that again, I think you want me to run 3dmark in those 3 cases

Lol, to see how much one can "cheat"? Hehe, OK I'll humor you

Vampyromaniac · Mar 29, 2003, 09:50 AM

OK, to answer your question:

1. 3089 at 150 Hz (Optimal rate at 1024 x 768)
2. same, 150Hz is the highest rate I can set at 1024 x 768
3. 3109 at 60 Hz (no interlaced rates available)

So in going from 150Hz to 60Hz, available memory bandwidth rose 5.55%, and 3dmark score rose 0.65%

Matth · Mar 29, 2003, 01:37 PM

Interesting...
http://www.ati.com/products/pc/radeon7200/

According to that, it's only SDRAM, not DDR, Although there is a DDR edition, apparently.

It seems the improvement definitely does not scale, and applies only to cards which are memory bandwidth limited - and I thought that was only the miserable GF2MX200 with a crippling 64 bit SDRAM memory bus.

I'm guessing that you have a 128 bit DDR memory bus at 166 on that card, as 166 sdram at 128 bit is comparable to the GF2MX.

So, no free lunch - the serious 3Dmark junkies would see even less gain.

Vampyromaniac · Mar 30, 2003, 05:50 AM

Quote:

Originally posted by Matth
I'm guessing that you have a 128 bit DDR memory bus at 166 on that card

Yet again, you are correct

Mine is actually an All-In-Wonder Radeon (32MB DDR), it uses the 7200 chip

Mar 28, 2003, 09:23 PM	#36
Matth Flash Banner Hater Join Date: Jun 2002 Location: UK Posts: 3,426 Rep Power: 93 System Specs	Just ran 2 tests on my Geforce 2MX (a Creative DDR version, at stock speeds 175/143 YES, they did use slow DDR, and it does perform worse than a good SDR model) At 75Hz, it scored 2248, while at 60Hz, it scored 2303** (3Dmark2001SE) A 2.4% difference, for actually somewhat less than that in bandwith terms. - I forgot - the bandwith saved need to be taken as a percentage of the bandwith LEFT after subtracting screen refresh usage from the total - even allowing for that, it's still more than 1:1 - presumably the overhead on the GPU is also reduced. Sure, I can gain the same and more by a RAM overclock - in fact, the score rises pretty well pro rata, so even though this is not the memory bandwidth crippled MX200, it's still memory bandwidth limited. If you refresh at 120Hz, while getting 60 FPS, the frame buffer is being read from the card RAM 120 times per second, and that is RAM bandwith not available for any other use. PS. Those scan rates are the limit of my LCD monitor, so while 60Hz looks totally solid (though vsync disabled tearing is actually far more objectionable than at 75), I can't go higher or lower to stretch the point. I'll lay the challenge - with a better card, and a monitor that can do it.... 1. Your normal refresh rate 2. The highest acceptable for your Monitor/card 3. 87i (Interlaced), equal to 43 full frames per second (or 60Hz if you can't) And if you CAN use 87i, and you gain any increase, is there anything in the results to show it - or is it a "not quite cheating" way to grab another fraction of a percent an climb the results table? Last edited by Matth; Mar 28, 2003 at 09:29 PM.

Mar 29, 2003, 09:50 AM	#38
Vampyromaniac confutatis maledictis Join Date: May 2002 Location: somewhere dark Posts: 5,974 Rep Power: 77 System Specs	Matth OK, to answer your question: 1. 3089 at 150 Hz (Optimal rate at 1024 x 768) 2. same, 150Hz is the highest rate I can set at 1024 x 768 3. 3109 at 60 Hz (no interlaced rates available) So in going from 150Hz to 60Hz, available memory bandwidth rose 5.55%, and 3dmark score rose 0.65% __________________ *Digitalis 3.3* Athlon 64 3000 // ASUS K8V SE Deluxe // 1024MB PC3200 (2-2-2-10 1T) ATI All-In-Wonder 9700 Pro // 20" Dell 2005FPW (DVI) M-Audio Revo 7.1 + Philips Acoustic Edge // Klipsch ProMedia 2.1 320/16 Western Digital WD3200KS + 120/8 Seagate 7200.7 NEC ND-3550A 16x DVD±RW + Lite-On 52x24x CD-RW Antec Sonata case // 480W Antec TruePower personal bests \|\| Aq'3: 46796 \| 3D'01: 20461 \| 3D'03: 6336 \| 3D'05: 2677 \| PC'04: 4605 \| PC'02: 7691,9092,1250

Mar 26, 2003, 05:49 PM	#32
~~shai~~ Banned Join Date: Jul 2002 Posts: 613 Rep Power: 0	well your a rich kid who pays for his own stuff...

Mar 29, 2003, 01:37 PM	#39
Matth Flash Banner Hater Join Date: Jun 2002 Location: UK Posts: 3,426 Rep Power: 93 System Specs	Interesting... http://www.ati.com/products/pc/radeon7200/ According to that, it's only SDRAM, not DDR, Although there is a DDR edition, apparently. It seems the improvement definitely does not scale, and applies only to cards which are memory bandwidth limited - and I thought that was only the miserable GF2MX200 with a crippling 64 bit SDRAM memory bus. I'm guessing that you have a 128 bit DDR memory bus at 166 on that card, as 166 sdram at 128 bit is comparable to the GF2MX. So, no free lunch - the serious 3Dmark junkies would see even less gain.