Forsa GeForce FX5600 < GeForce 3 Ti200.. Don't make the same mistake I did..

[chsh1ca]

Okay, I recently was posting about picking up an MSI 5900XT because it fit my budget, but something came up, and saw my budget halved, but I really need dual monitor outs for day to day use, so I opted to go with instead a Forsa GeForce FX5600 XT, which according to their spec sheets offers 8.8GB/s of memory bandwidth. After looking at the fill rate results, I think this is total crap. After double-checking the page, I see this is listed as being either 64 or 128bit bus width, and I have a sneaky suspicion this card has a 64-bit bus. What doesn't make sense to me is why have two versions of the same card, with one having half the performance?
Upon receipt of the card, I tossed it into my box and tried out 3DM03 to compare my results with my earlier run on the GeForce 3 Ti200 128MB I had. Needless to say I am disappointed, and contemplating returning the card. For the price, I could probably get a dual out Ti4200 or a used Radeon 9600.

Wins on a per-test basis are highlighted in bold.
GeForce 3 Ti200 128MB Results | GeForce FX5600XT 128MB Results | GeForce 2 Ti 64MB

Res: 1024x768x32@85hz, AA/AF off.
Overall: 1013 | 1344 | 231
GT1: 58.0 | 55.2 | 28.9
GT2: 8.0 | 8.6 | N/A
GT3: 6.3 | 7.3 | N/A
GT4: N/A | 7.2 | N/A

CPU Score: 378 | 341 (WTF??) | N/A
CT1: 37.3 | 34.4 | N/A
CT2: 7.6 | 6.7 | N/A
FR (ST): 464.3MTexels/s | 321.6MTexels/s | 396.9MTexels/s
FT (MT): 1022.0MTexels/s | 674.0MTexels/s | 742.2MTexels/s

VS: 2.2 | 1.9 | N/A
PS2.0: N/A | 8.6 | N/A
RagTroll: 2.6 | 5.9 | N/A

SND(0): 24.0 | 23.3 | 16.5
SND(24): 20.5 | 20.1 | 14.4
SND(60): N/A | N/A | N/A

==== Now for notes about game benchmarks ====

Tactical Ops (Mod for Unreal Tournament running in OpenGL mode):
I took some averages of both inconsistent online play, as well as the demo for a clan war I participated in and replayed on each.
Res: 1280x1024x16 (32 bit colour doesn't make a noticeable texture difference and incurs a performance hit on both cards)
Comparison:
GF3 Avg | GFFX Avg:
Online play: 126FPS | 87FPS | 95FPS
Demo: 124.7 | 88FPS | 94FPS

That's what I've benched so far, since all I mainly play is Tactical Ops (though I am playing quite a bit of Desert Combat and BattleField 1942 lately).

Subjective observations: The FX dips below 60FPS quite regularly, which was never a problem in some of the older games I play with the GF3. At the framerate the FX is pumping out, I could only equal it in TO on the GF3 at 1600x1200.

GeForce 3 results were done with the 45.23 Detonators, and the GeForce FX results were using the 53.04 Detonators.
As I said, I'm highly doubtful that this card has a 128-bit memory bus given its poor performance, and I am contemplating returning it since it is way less than my expectations of it, combined with how misleading their site is. According to them, the fill rate of the card is 1.3billion texels/s. This card barely manages half that.

System is:
Athlon XP 2000+
512MB PC2700 DDR RAM
Asus A7V8X-X (onboard sound disabled, onboard BCM-4401 NIC enabled)
MSI GeForce 3 Ti200 128MB/Forsa GeForce FX 5600 XT
Creative SoundBlaster Live!
3Com 3c905B-TX 10/100Mb NIC
Maxtor 7200RPM/ATA-100/20.4GB IDE HDD
Fujitsu 7200RPM/ATA-100/40GB IDE HDD
Pioneer 16x IDE DVD-ROM
Samsung 8x4x32 IDE CD-RW

Note: Windows only actually uses a 10GB partition on the Maxtor, so the Fujitsu doesn't enter the equation.

Pros:
- Really cheap (now I know why, I had originally thought it had something to do with the FX5700s being out)
- Compared to my previous intention of doing a dual monitor setup using an S3 ViRGE 3D aside my GeForce 3 this is a marked improvement. Manufacturer does not supply DVI-I to VGA connector with the card, so it's a good thing I had one lying around.
- Will actually run some of the DX9 tests they GF3 wouldn't (is this actually a pro? )

Cons:
- Performs barely better than a GeForce FX5200 which is absolutely nowhere near what my expectations were.
- Manufacturer's site is misleading, and I'm tempted to say borders on false advertising.

For those who are too lazy to surf the site, here's the specs they list:
- Graphic Core : 256-bit
- Core Speed : 235MHz
- Memory Configuration : 64/128-bit 128MB / 256MB DDR
- Memory Speed : 400 MHz
- Memory Bandwidth (GB/Sec) : 8.8
- Bus Interface : AGP 8X
- Pixels per clock (peak) : 4
- Veticles/sec. : 81 million
- Fill Rate : 1.3 billion
- RAMIDACs (MHz) : 400
- Max Resolution : 2048x1536 @ 75Hz
- Monitor Support , Analog Monitor , HDB-15 connector , TV-out, S-Video , Digital Flat Pane l, DVI-I

EDIT: Added 3DM03 results for the GF2Ti I have in the box right now. Will add TO results shortly.
Note the fill rate on this 64MB GeForce 2 Ti is still higher than the GeForce FX5600's.
EDIT2: Okay, so it performs better at TO than the FX does, the extra 10FPS is the difference between enjoyable play @1280x1024x16 and having to switch down to 1024x768x16.

 

[BlvdKing]

That's sad that it can't even beat a GF3 Ti 200. You would definately be better off with an ATI 9600.

 

[Blastman]

Ug, ? ? those 5600XT?s are ? really really really ?? bad cards. :disgust:

Consider that a 9600SE (64 bit memory bus) in this elitebastards review was about 2x as fast as the 5600XT it was pitted against. The 5600XT may be close to the ti200 in 3Dmark2003 but that?s because the ti200 chokes on DX9. In regular gaming the 5600XT is probably going to be substantially slower than the ti200.

In that elite review the 9600SE actually surprised me with how well it performed. They were able to run some AA/AF on a lot of the games and still achieve acceptable performance. It ground that 5600XT into the dirt.

The Sapphire 9600SE 128MB is $67 at Newegg. The Sapphire 9600 128MB is $95

I would go for the 9600 for $95, it?s a lot faster than the 9600SE and can be overclocked very nicely too, just bolt a little fan on the heatsink and the card should be close to a 9600pro in performance. AkumaX got his 9600 up to ? 455/255 DDR...(from stock 325/200).

 

[modedepe]

Well it's not surprising it won't be a gf3. The 5600 is already a weak card, and giving it a 64bit bus is really going to kill it.

Also, I think maybe they're taking into account z compression in their memory bandwidth figures. I'm not sure how else they'd get 8.8GB/s.

 

[kylebisme]

Originally posted by: chsh1ca
Needless to say I am disappointed, and contemplating returning the card. For the price, I could probably get a dual out Ti4200 or a used Radeon 9600.

contmeplating?!? do it!

 

[chsh1ca]

Originally posted by: TheSnowman

contmeplating?!? do it!

Yeah, I sent the email about it to the vendor earlier today. Hopefully I'll have a response tomorrow morning.

Originally posted by: Blastman

The Sapphire 9600SE 128MB is $67 at Newegg. The Sapphire 9600 128MB is $95

I would go for the 9600 for $95, it?s a lot faster than the 9600SE and can be overclocked very nicely too, just bolt a little fan on the heatsink and the card should be close to a 9600pro in performance. AkumaX got his 9600 up to ? 455/255 DDR...(from stock 325/200).

Unfortunately due to luxury taxes and duty and such up here, the cheapest I can get a R9600 for is a PowerColor R9600Pro for $198 CAD or a Gigabyte R9600SE for $125CAD. If NewEgg shipped to canada, I could get a better deal on exchange, but the duty would make it more expensive, so either way... I recall a lot of quality-related problems with powercolor, so I doubt I would want to go with a powercolor card.

Originall posted by: modedepe

Well it's not surprising it won't be a gf3. The 5600 is already a weak card, and giving it a 64bit bus is really going to kill it.
Also, I think maybe they're taking into account z compression in their memory bandwidth figures. I'm not sure how else they'd get 8.8GB/s.

I know, I did the math, and I think they can't add. 128bit@400MHz = 6.4GB/s, 64bit@400MHz = 3.2GB/s. I think it is a huge error on their part, and I'm sorry I overlooked it since at the very least I will be out of shipping costs on this bloody thing. I think they just copied the specs for the FX5600, which would actually be 128bit@500MHz (8.0GB/s), which still isn't correct. It'd have to be 128bit@550MHz to hit 8.8GB/s, how you misplace 150MHz and/or half the bus width in a spec sheet as the manufacturer, I have no idea...

 

[Blastman]

NCIX? has the Sapphire 9600 for $142 (+ $11 shipping). CAD.

ATIC.. has the 9600 for $132.

If you like NCIX get them to price match ATIC and that will pay shipping costs.

 

[Blastman]

These guys are pretty good too ? dangeo?

You could try searching using ?.
pricenetwork ? very good search engine to find hardware prices here in Canada.

 

[chsh1ca]

Thanks for the links, but it looks like I won't be seeing my money back. The guy replied to me and said basically "take it up with the manufacturer", so I've sent that email off asking them to rectify the situation. Either they will refund me and I will end up with a Radeon 9600 or they will simply upgrade my model (according to the vendor, that's what they would do in this instance) to a proper one, which is either an FX5600 or an FX5600 XT Pro which is only available in 128MB/128bit bus version...

EDIT: Just send off the info at 11:40PM EST on the 26th, got a response from forsa's tech support department in less than two minutes. Must be business hours in Hong Kong.

 

[MichaelZ]

Originally posted by: chsh1ca
FX5600 XT Pro

...

 

[yhelothar]

9600 isn't much faster than geforce3 either. I went from geforce3 to 9500np, and they performed pretty much identically.

 

[Jeff7181]

People underestimate GeForce3's just because they're two generations old. Their biggest weakness compared to today's cards is of course DX9, and AA and AF. GeForce3's are what I would now consider budget cards... about a year and a half ago I bought my brother a GeForce3 Ti200 for $75 and he plays Call of Duty with it now without any AA and AF. I played it when I was visiting, and to be honest, it plays just fine... I didn't measure the FPS, but I would guess no less than 50. That was at 1024x768 too.

 

[chsh1ca]

Originally posted by: virtualgames0
9600 isn't much faster than geforce3 either. I went from geforce3 to 9500np, and they performed pretty much identically.

You're missing the point. I am fine with equal performance in DX8, but dual outs and actually decent performance in DX9 (what I was looking for). I'm not necessarily looking for a card that will annihilate my GF3's speeds in old games card. What I am looking for is a card that won't halve to quarter my performance. I feel pretty much misled because the numbers on their site are all screwed, and like an idiot I suppose I thought that the manufacturer would be honest. Their response was to ask me for the information on the barcode on the card, and when I supplied it they told me it was a version that had a 64-bit memory bus, and that yes, compared to a GeForce 3 Ti200, it would be slow...

 

[rancherlee]

On thing I notice about the 9600 PRO that I have is it was actually slightly slower than my 8500 retail in games are CPU limited (like UT2K3) as both have 128meg ram. Also the first 3 tests on 3dmark 2k1 are mostly CPU limited with newer cards, my 8500 runs neck in neck with the 9600 pro. BUT on Nature which is more graphics card limited the 9600 pro walks the 8500 by a good 20% Once overclocked the 9600 pro (490core/730mem) it will beat the 8500 overclocked (306/612) in every test by ~10% and 40% in nature. 3dmark 2k3 is a different story, the 9600pro over DOUBLES the 8500 in most all the tests.

Unreal tournement 2k3 the 8500/9600pro run the same frame rates all the way up to 1024x786x32 (DEAD even) once I turn on 2xAA/8xAF the 8500 becomes a bit choppy on big maps and the 9600pro runs smooth, 1280x1024x32 the 9600 pro get slightly choppy with 2xAA/8xAF on big maps but is barely noticable, the 8500 is unplayable at that setting.

The 8500 and compairable to the Geforce 3 TI200 for the most part.

 

[stnicralisk]

The fx5600se/xt is a budget card I am not sure what you expected.

One thing I will ask is.. why are you posting your cpu test marks and your sound test marks... and instead of using a mod that runs something in opengl.. why not use an actual opengl game might have less problems.

 

[chsh1ca]

Originally posted by: stnicralisk
The fx5600se/xt is a budget card I am not sure what you expected.

I thought I made it rather clear. Even 'normal' XTs seem to have just slower ram (400MHz vs 500MHz), and according to their numbers, it should perform about on par with an FX5600. It should not HALVE my performance given what they indicated the card did.

One thing I will ask is.. why are you posting your cpu test marks and your sound test marks... and instead of using a mod that runs something in opengl.. why not use an actual opengl game might have less problems.

Well for one, the CPU and Sound tests were different between the two cards. For two, I fail to see how using an Unreal Tournament mod is going to provide less performance or create "problems" when benchmarking it relative to another card. The only really half-recent games I have are BattleField 1942 (which is pretty difficult/impossible to benchmark) and Warcraft III. I also own a copy of Planetside, but I don't play it anymore, too many bugs. I do intend on downloading the UT2004 demo, but it's a moot point at this point in time. The card I was sold gets nowhere near the performance their specs indicate, so I'm returning it either for my cash back or for a card that does do what the specs say.

 

[WobbleWobble]

You're very unlikely to get a full refund simply because the performance is not what you expected. I'm sure they accept returns with a 15% restocking fee.

 

[chsh1ca]

Originally posted by: WobbleWobble
You're very unlikely to get a full refund simply because the performance is not what you expected. I'm sure they accept returns with a 15% restocking fee.

If I expected a certain level of performance it was based on the misleading tech specs on their site, which is why I believe I, and anyone else who got this 64-bit version of the 128MB card, are entitled to full refunds. According to their technical support department, they have a 128bit version of the 128MB card with the exact same packaging and naming. The only distinguishing feature between then appears to be something on the barcode label. If say Gainward started making FX5900s with half the mem. bus of other cards of the same line, without making a clearly identifiable difference, people would scream about it, that's all I'm doing here. I doubt there is a restocking fee anyway since the seller told me to take it up with the manufacturer.

 

[WobbleWobble]

You have a very valid point. But I think that they will claim that the memory bus is not the important selling point of the card, only the FX5600 chip and the 128MB of memory.

But good luck! I hope you're able to get something done.

Hopefully you can get them on their advertised memory bandwidth

 

[chsh1ca]

Okay, spoke to the seller, he insists it isn't his problem, as he referred to the manufacturer's specs, lol. I'd recommend not buying any FX5600 line card from Forsa, since the packaging itself is incorrect. Check out these pics which are the box this card came in:

Box Front
Box Back
Card Front

EDIT: Added link to card pic.

 

[Blastman]

What a pile of garbage on that box. They list the 5600Ultra specs on the box. And 128bit ? the drivers for the 5600 never use that precision. :disgust:

 

[chsh1ca]

Indeed. I'm going to be contacting the consumer protection people here in Canada regarding this particular product. I think other people who have purchased this card probably should get a similar warning about its performance and basically the misleading and utterly false advertising done with the card. I repackaged the card up yesterday when I sent the initial email, so now I'm using my brother's PINE GeForce 2 Ti 64MB TV (I already swapped my GF3 Ti200 into his machine, so I can live with th GF2 for a bit). For kicks I am going to run the same tests with this card to see how it performs relative to the FX I just had in it.

 

[chsh1ca]

Update: Spoke to the consumer protection people, they say regardless of who advertised what, by selling the product in canada the vendor is solely responsible for ensuring the accuracy of the information dispensed during the sale. So I sent of an email stating this (with phone # reference and appropriate law linkified), and basically asked the guy if it was really worth his time to argue about it and end up having to replace the thing anyway.

I must stress this again, DO NOT BUY A FORSA GEFORCE FX 5600! They seem to be incredibly confused about which card is which. They list specs for the "XT" on their site as being almost identical to that of their regular FX5600, and on the box it comes in they list the specs for the FX5600 Ultra. At any rate, I will keep this thread updated in case anyone else has an issue with this card.

 

[bandana163]

8.8GB advertised memory bandwidth?
Impossible. Even the fillrate in 3DM03 is below 700MTexel/sec, which is plain sad.
Memory bandwidth theoretical maximum values without further "optimalizations":
128 bit +400MHz = 6.4GB/s
64 bit + 400MHz = 3.2GB/s
That packaging is surreal, it might give the user a reason to sue the manufacturer for misleading him and not selling the advertised product.

 

[chsh1ca]

Update again, I sent a final email to the vendor, and am pursuing this on a legal front. I would like to post the vendor's name and a summary version of this thread somewhere, but I don't think here's the place to do it. Any ideas? My initial thoughts would be to post this info to the "Bad Traders" thread on FS/FT, but I can't really say that since this wasn't an AT trader, but rather a company who regularly puts up stuff on eBay. In addition to a warning about this vendor, this thread should probably also serve as a warning about avoiding FORSA cards, since they seem to be equally deceptive with the way their advertising is done. I appreciate input.

Edit: I've also filled out an eBay fraud report regarding the item as was sold (which can be found here). As you can tell, the vendor himself copied the specs into the item description, and as I've documented above, the thing doesn't perform anywhere near as well as he advertised.

Any long-time ebayers able to lend me some advice on whether I should contact the other people who bought this item from the guy?