NVIDIA nForce Chipset Review. Part II: Graphics, IDE, Sound

[NFS4]

http://www.xbitlabs.com/mainboards/nforce-2/

 

[Adul]

cool, though I think I want to try out the nforce without video on it.

 

[Soccerman]

I don't know why people were complaining about Via IDE performance (perhaps this isn't quite the situation they were testing under perhaps) but I don't see any sign of it here! thats for sure...

they should have tried to test the Ethernet :-( but oh well..

Gfx was ok, my Radeon beats it though

ahh! sound! this is the part that I want to hear about! guess what? it's not that great! doh!

really, it has good CPU usage, but apparently (in UT) low quality. could be a driver bug, who knows..

conclusion: the nForce might be a good server board.. MIGHT. it's sortof an oddball platform IMHO..

 

[AGodspeed]

cool, though I think I want to try out the nforce without video on it.

Exactly my sentiments Adul.

Judging from the benchmarks of XBit's review, the nForce APU is equal if not better than the Audigy in several key areas. However, I was quite shocked to hear him say that the sound quality wasn't up there with the best, mostly because every review I've read has raved about the sound quality. He seems to blame this on the reference board, and I would have to agree with XBit on that one, since my MSI K7N420 Pro sounded great the last time I played CS and just plain music. I guess motherboard layout is important for the APU or something.

I'll be anticipating a $100 415-D nForce board in the next month or two.

 

[NFS4]

<< cool, though I think I want to try out the nforce without video on it.

Exactly my sentiments Adul. Judging from the benchmarks of XBit's review, the nForce APU is equal if not better than the Audigy in several key areas. However, I was quite shocked to hear him say that the sound quality wasn't up there with the best. He seems to blame this on the reference board, and I would have to agree with XBit on that one, since my MSI K7N420 Pro sounded great the last time I played CS and just plain music. I guess motherboard layout is important for the APU or something.

I'll be anticipating a $100 415-D nForce board in the next month or two. >>


Yeah, I wouldn't mind having an nForce in my system for ~$100...IF THEY FIX THE MEMORY BUG I'd pick NVIDIA over VIA, SiS, or ALi if it came down to it.

 

[AA0]

Low quality sound is most likely the expensive of your cpu cycles saved. Its not unexpected, the headphone thing is really dissapointing, lots of people use those, they shouldn't have cut corners.

And the IDE performance is on par with VIA, guess some people last week put their foot in their mouth, eh?

I'd like to see a comment about 2D quality on the integrated graphics.

 

[Soccerman]

what's up with this 415-D board? is that the same nForce as the 420 without onboard video included? I could probably hazard a guess that the performance wouldn't be any better than the 420 /w onboard video disabled if that's the case... why would u want it? cause of the onboard audio?

 

[robg1701]

...cause its not Via.

 

[AGodspeed]

why would u want it? cause of the onboard audio?

Of course, why not? It's not like I'm going to go out and buy an SB Live! card if the nForce APU takes it apart. And I'm certainly not going to buy a $50 or $60 Audigy (with poor WinXP drivers) or Santa Cruz if the nForce APU is quite competitive with equal to or better drivers.

Also, the 415-D won't have GeF2 MX graphics, and as Aceshardware has indicated, this should bode well for overclockers. Not only that, but the 415-D is supposed to start around $100, which IMO is something to get excited about.

 

[BD2003]

Hmm, sell my current MB for $75, my Audigy for $60, my ethernet for $15, my IDE controller for $25 (via is junk, need it only for via) my modem for $15....

Buy Nforce, get superior quality in almost all respects...and save $90. I want one!

 

[jpprod]

Low quality sound is most likely the expensive of your cpu cycles saved.

It's more likely the sound quality was low because of a cheap DA-conventer chip used, or poor quality connectors on the motherboard. It's basically the same reason the 2D of some nVidia boards is bad: the hardware basically rocks while actual quality does not, precious pennies has been saved in the interface between digital and analogue worlds.

I'll spare my judgement on the nForce audio until I'll see a test where the sound quality is compared to Live series using a digital audio out connector.

 

[AGodspeed]

Just so everyone knows, this is a reference board, not a top tier board like MSI or ASUS. That might be the culprit, but who knows.

Perhaps XBit missed Nvidia's latest Unified nForce Drivers released just 2 days ago.

 

[Soccerman]

...cause its not Via.

I'm sick of all this VIa bashing. first of all, I don't see any IDE performance problems in that comparison.

second of all, if you have Creative Labs SBLive! on a 686B, then sorry to say but your out of luck. besides I hate Creative for what they did to Aureal. the Audigy is the only real gamers card now, and I don't trust it to work enough to slap it in a 686B mobo..

third I don't know of any other issues that haven't or can't be solved with a driver fix.

HOWEVER

Of course, why not? It's not like I'm going to go out and buy an SB Live! card if the nForce APU takes it apart. And I'm certainly not going to buy a $50 or $60 Audigy (with poor WinXP drivers) or Santa Cruz if the nForce APU is quite competitive with equal to or better drivers.

did you actually read the review? notice how they were testing in XP? basically they didn't have proper drivers for the SBLive! card (AND ADMITTED IT), so those results are probably invalid.

really, how bad are the Audigy drivers? I haven't hear people complaining about them. the only problems I hear about the Audigy are some people with AMD chipsets having crackling (WTF is it with Creative and crackling?).

anywho, right now becuase I hate Creative (now they have an Extigy which uses another device I happen to dislike: USB!), I would get a Hercules GTXP, not a Santa Cruz.. u get the same functionality of the nForce APU (without the DD encoding though). at least that way I don't have to worry about audio quality on my motherboard (sheesh, you'd think that we didn't have enough to worry about when making our mobo decisions).

Hmm, sell my current MB for $75, my Audigy for $60, my ethernet for $15, my IDE controller for $25 (via is junk, need it only for via) my modem for $15....

Via is junk?? show me your proof? I'm running a 686B which means ATA100, and have no problems whatsoever.

It's basically the same reason the 2D of some nVidia boards is bad: the hardware basically rocks while actual quality does not, precious pennies has been saved in the interface between digital and analogue worlds.

the hardware is what's making the 2D bad it's all hardware, but going nVidia means you have to do extra work finding a card in your price range made by a certain company (leadtek I think it is).

 

[BD2003]

<< Via is junk?? show me your proof? I'm running a 686B which means ATA100, and have no problems whatsoever. >>



Consider yourself lucky then. Besides, I'm not railing against via telling people not to buy it, its just Ive had more than enough problems with VIA chipsets. DMA not working properly, slow accesses, corruption, you name it. Havent had a problem since I plugged the Promise Ultra 100 in. I dont need any proof for myself other than my own experiences.

As far as the audio on the nforce goes, I hope they use the worst possible D/A they can find. That way I save a few bucks while I digitally hook it up to my 5.1 system...

 

[Gstanfor]

You have to wonder why x-bit labs is bothering to test a reference board at this stage of the game ???.

nfs4:


<< IF THEY FIX THE MEMORY BUG >>

I don't know what memory bug you refer to. If you use approved memory modules in the correct slots as indicated by motherboard maker you will have NO memory problems. The fact the nforce even supports 3 DDR dimms is more than most competing chipsets can manage in the market place.

When is anand going to review the MSI or ASUS production boards BTW?

sound issues:
I've encountered none with the latest drivers on the K7N420 Pro. It is one of the nicest soundcards I've ever listened to. As others have suggested I think this is a result of testing a reference board.

AA0:


<< And the IDE performance is on par with VIA, guess some people last week put their foot in their mouth, eh? >>


I can assure you the IDE performance is NOT VIA like, at least not on the MSI. What do you and soccerman average when ghosting a partition (mb/s)? on my nforce I average 91-97 mb/s, on K7T Turbo 58-69 mb/s. Note the K7T Turbo has the same southbridge as the VIA DDR boards. Both controllers puport to be ATA100.

Greg

 

[Soccerman]

your hard drive can't even sustain 90 unless you're doing IDE/SCSI RAID..

58-69 sounds a bit too high too.. perhaps next generation drives can hit that but I don't think todays can (discounting bursts from onboard cache).

 

[Gstanfor]

I'm only reporting what Ghost itself reports (under pure, floppy booted MS-DOS), nocache, nodriver conditions.

Greg

 

[robg1701]

Soccerman, you did see the smilie, right ? >>>>>>>>

It was meant as a little bit of a light hearted statement, and im not an active 'via basher', i dont go round putting em down, ok....


BUT......I do admit I persoanlly dont really want to buy a Via chipset because of all the problems people have had, I have friends who are experiencing good and bad things with systems containing Via chips...one is a KT133A, one is a 760+686b combo....the latter being the one that is working fine, though in saying that its had its quirks with DMA in the past. I am stilling running an old Slot A Athlon 500 on an all-amd based MSI K7 Pro, and quite happy with it, never had a problem. Im going to be upgrading in roughly 4 months, and while previously i was defiantely not gonna get a via chipset, the KT266A was certainly winning me over...until the latest PCI bus sh$t hit the fan...sure the Live! may abuse the PCI bus, but maybe the Via - LIVE! thing is not all Creatives fault (and maybe it is, i dont know, neither do you really...) but i do know that i am almost scared to buy a Via chipset mobo. If i had one and it worked great, so be it, but from where im standing its unnerving. It will be an Athlon system i get, but Via is last in line at this moment in time(ok, theyre ahead of ALi ). Bring on the nForces (maybe Crush 17/18 derivatives ?), Sis745's, hell even the ATi A-Whateva's i say....and ive actually thought about a 760MPX once theyve fixed the 768 up...ALL-AMD, cant beat it baby....I believe you (I may be wrong, but I think ive seen you say you got a dually 760MP setup) know this ?

 

[InsaneTBONE]

The memory placement or type is not a bug. What nvidia did was quite purposeful.

 

[robg1701]

Gstanfor, thats not MB/s, ghost shows MB/min figures.....

 

[Gstanfor]

Heh. You are correct.

With regard to the Pro/Anti VIA part of this and other threads, personally I thank VIA for supporting the Athlon early on when almost noone else would. However I don't thank them for then sitting on their success and default postion of dominance in the Socket A market, and not offering users true performance until forced to by competitors. If SiS and nVidia had not announced their chipsets when they did we would still be using the original KT266.

Greg

 

[Soccerman]

ya I saw the

the only thing nForce has going for it is the DD encoding (which I woudn't want anyway), and the fact that the audio, and ethernet is integrated..

now as for actually testing my hard drive, I'd have to download a program that isn't Sisoft to test it (I don't trust Sisoft Sandra at all), and even then I don't know how my CPU speed, amount of RAM, or even video card would affect what I get.

if you actually trust the scores your getting from ghost then I think you're probably using compression or something like that. are the hard drives the exact same type in both situations?

however that still wouldn't allow for a 90 meg/second transfer rate cause the HD can't write that fast.. I believe I was getting some unreal numbers when I used ghost with my CDRW drive (put the image on a CD, restore it) but that was so long ago I can't remember..

DMA not working properly? is that in windows? I'd like to see a before/after benchmark with DMA checked and unchecked (if you can check it in the first place). anywho, AFAIK DMA only really affects CPU usage.

no I don't have a Dual athlon mobo but that sure would be nice

blah, I don't want to talk anymore.. to each his/her own..

 

[robg1701]

Well, as it happens, i did do a benchmark of sorts on the guys system...i sat there for 4 HOURS one day ripping every single item on the Star Wars Episode 1 DVD's for a 4CD 'super rip' including every single item avalable in the set (even the easter eggs) and replica Flash menus i spent a couple weeks on....it took so long because it couldnt get past 1.4X speed and tehre are so many individual items to rip, not just one big movie....DMA was enabled, and acording to windows, working...checking himself with other DVDs than mine in order confirmed this, the cpu would be under heavy use and low ripping speeds were achieved...moving the drives around the channels was the only way to get it sorted, and it tested it with other DVDs since, though not my Ep 1 discs...i however have ripped tham at 6x on my 10x drive (his is a 16x drive, 1.3GHz athlon vs 500Mhz Athlon) so i doubt the discs were at fault

 

[NFS4]

<< first of all, I don't see any IDE performance problems in that comparison. >>


That's a bunch of crap that Pabster has been throwing around for the lord knows how long.


<< I don't know what memory bug you refer to. If you use approved memory modules in the correct slots as indicated by motherboard maker you will have NO memory problems. The fact the nforce even supports 3 DDR dimms is more than most competing chipsets can manage in the market place. >>


That so called Super Stability crap:

If it weren't a bug, they would have been straight forward with the "feature" and let us know from the beginning.

http://forums.anandtech.com/messageview.cfm?catid=27&threadid=647320&highlight_key=y&keyword1=super

 

[AA0]

It is most definitely a bug, and not an option. Anyone saying otherwise is just an nvidiot.

No company that is sane would use slot B as the super stability slot.

IMO, the chipset is crap. It isn't ready to be released, as they never got the memory timing right, which is a difficult task for DDR. When you combine it with the low quality sound (don't blame it on sub-standard parts, its their own reference board), and VIA IDE performance, there is really no reason to go with it. The ethernet controller is the only nice thing, and that comes on most boards as options.