Slashdot Mirror
nForce2 Preview
An anonymous submitter writes "I noticed that a review of NVIDIA's nForce2 chipset has been posted here. From what I can gather the chipset contains two 10/100 ethernet controllers, six USB 2.0 ports, UltraATA133 support, three 1394 ports, five PCI slots, and an integrated GeForce4 MX core including NVIDIA's nView technology and a TV Tuner." Tom's Hardware and NVNews also have looks at it.
This chipset is designed to be used in OEM boards for good performance and enthusiast customers...not servers. I can't think of any legitimate use for *two* ethernet controllers other than in a broader network application (Firewall, for instance).
Perhaps you could use it to make a really stupid sort of bus network for LAN parties using nothing but crossover cables, but that's such a silly idea (performance/configuration issues) that it's probably true...
"Mod, mod, mod...and another troll bites the dust."
for a home with more than one computer with a cable modem this makes perfect sense. For a couple dollars more, it would be stupid not to...
due to the newfound explosive nature of mice, I'm going to skip out on buying hardware for the time being.
Not sure you are meant to be as a flame or not, but yes you do need a CPU for this, an Athlon CPU to be exact.
IIRC, I don't think NVIDIA has licence for making Pentium 4 chipset, so they are pretty much stuck with AMD's processors for now (I think they are making chipset for Hammer as well)
I think nForce 2 is great but watch out for ATi's new chipset as well. (I smell a chipset super pricewar in the distant future!)
kawai
On the topic of current Nvidia cards:
.plan:
Do not buy a GeForce4-MX for Doom.
Nvidia has really made a mess of the naming conventions here. I always
thought it was bad enough that GF2 was just a speed bumped GF1, while GF3 had
significant architectural improvements over GF2. I expected GF4 to be the
speed bumped GF3, but calling the NV17 GF4-MX really sucks.
GF4-MX will still run Doom properly, but it will be using the NV10 codepath
with only two texture units and no vertex shaders. A GF3 or 8500 will be
much better performers. The GF4-MX may still be the card of choice for many
people depending on pricing, especially considering that many games won't use
four textures and vertex programs, but damn, I wish they had named it
something else.
(all this comes from carmack's
http://webdog.org/plans/1/ )
It seems nvidia is going the same road as intel and sis with their cheap video-on-board motherboard. All of them sucked! Good luck!
I tried to read that artical, but the flashing "you've won" advertizement was making my eyes bleed.
autopr0n is like, down and stuff.
*snip*
integrated GeForce4 MX core
*snip*
My question is this: who are they trying to sell this to? Not gamers, since a GeForce4 MX is a stripped down, cheaper version of the real powerhouse GeForce4 TI, which is the new bar for nVidia cards. Obviously, they're not selling this to power users who build their computers piecemeal, because, well... an integrated board by definition defeats that purpose.
Granted, gamers isn't where the money is. The money is in getting someone like Dell or Gateway to use this board in their corporate lease computers. By convincing big manufacturers that the overall cost of making a computer is lower by buying one big all-in-one solution board, they hope to break into new markets.
*snip (from review)*
aimed not only towards the high end but the mainstream
*snip*
There you have it. It's a great product, but if you're a typical slashdotter, you're probably not going to care because:
a) nVidia Linux support has been a bit shoddy (IMHO - although the fact that they have drivers at all is a positive note)
b) it's not high end - it's a glorified GeForce2
c) it's integrated, meaning hard to replace if something goes out and not customizable
Plans for nForce 4 (still some time in the future) include an embedded version of Java and/or Internet Explorer
Ñ'
I'm not sure whether to feel juiced about the fact that the nForce has USB 2.0, Ethernet, Firewire, TV Tuner (!!) and a bunch of PCI slots built in automatically, or unhappy that they've paired all of these great features with what amounts to a budget on-chip video card.
Intel has always been a little hostile to anyone else producing chipsets. I believe with the Pentium IV, Intel has forbidden any third parties from producing chipsets. They went after VIA hard over this issue. I don't think Nvidia wants to deal with that sort of bullshit.
I also believe that Nvidia realized something with their Geforce 3, the damn thing was more advanced than many of the CPUs available at the time.
We are at an interesting point in computer history here, with graphics chips being as advanced or more so than the CPU, it is only natural that the two be brought closer together.
AMD and Nvidia seem to be doing that, while Intel is not really paying attention.
I don't read or respond to AC posts
What, no kitchen sink?
What? No open AGPx4 (or x8) port for my ATI 8500? ;)
John Maynard Keynes: "When the facts change, I change my mind. What do you do?"
And then the board would cost $200 more. You want super duper graphics for gaming, then buy a real GeForce card. I'm sure they thought about this for 10 seconds and came to the realization that the increased cost would decrease their market significantly.
The chipset contains 5 PCI slots? I'm impressed. I had to buy a motherboard to get that.
If you aren't part of the solution, there is good money to be made prolonging the problem
Or does anybody else think these are just press releases, not reviews, as listed??
Mozilla
nVidia's marketing department should be ashamed; the name of this piece of hardware is blatantly misleading. Every other "mx" version of their cards contained the same featureset of it's GeForceX line, but had slower/less memory.
The GeForce4mx, on the other hand, is missing the priciple feature of the GeForce3, that being hardware vertex and pixel shader support. The GeForce4mx is basically a really fast GeForce2. It's a sham.
It screws developers (no longer can we say "GeForce3 and up", we have to qualify by specifically excluding the GeForce4mx). It screws customers by making them think they get a better card than they are. It's just bad all around.
When I talked to an nVidia rep at this year's GDC he acknowledged it's hatefulness and gave the impression that it would be going away shortly. Given the number of these cards I see in stores and this announcement, I'm starting to doubt him.
Note to nVidia: when your marketing department starts screwing developers and customers, we developers stop wanting to support your cards. You've been at the head of the pack for a while now. Crap like this isn't how to stay there.
The current nForce is available in small form factor mobos like the Abit NV7M. I'd be very interested in a tiny version of nForce2.
If you plan to use TwinBank (6GB/sec system bandwidth!) you only need/want two DIMM slots. With video, audio, network, firewire, usb2, etc, all built-in, you hardly need the PCI slots at all.
Fewer components should also mean lower power consumption, which means fewer/slower fans, which means blissful quiet computing. Hopefully.
Also, any word on the rumored Shuttle SS41 yet?
Seems you skimped a little too much on the motherboard, as it's not even on the parts list. Perhaps if you had the crucial device connecting your CPU, RAM, graphics and sound cards together, you might get a little performance out of them. ;-)
but seriously... Not sure what bus speed the 1.0GHz Athlon does (I think it's 200MHz), but more than anything else, it's bus speed that matters - of course, you need a board that can support high bus speeds to maximise performance, which, if you have DDR RAM, you should have. You'll still be needing a CPU with 266MHz bus to maximise performance from your current setup, though. Could be time to get an Athlon 1.4GHz, at least.
Like car accidents, most hardware problems are due to driver error.
Is it me or are the lackies being hired at Tom's are getting dummer and dummer with time? I don't have time to run down the entire list of inaccuracies and errors in the article, but according to them DDR400 "corresponds to a performance level that SDRAM could only achieve at 400 MHz," and best of all, Nvidia was "Founded in 1997 by a handful of ex-SGI employees."
Anandtech's article This one is much more than just a "breif" overview...it is meaty :)
Unstable Apps: Our Android Apps Don't Suck
According to the NVNews article, they have a reference motherboard?
That's not what it says. If you read closely, you'll see that they tried to simulate the performance of an nForce2 mainboard by using an nForce motherboard with an underclocked GeForce4MX 460.
...you would realize they releasing a verion with integrated video and a version without.
- Toby
Does anyone know if the BIOSes for the nForce (and probably the nForce2 if the nForce has it) have PXE support for network booting?
ANY provision for network booting?
netbooting would make these into killer thin-client motherboards...
retrorocket.o not found, launch anyway?
Intel has always been a little hostile to anyone else producing chipsets. I believe with the Pentium IV, Intel has forbidden any third parties from producing chipsets. They went after VIA hard over this issue.
Intel actually licenses the IP necessary to design chipsets for the Pentium IV. The reason that they went after VIA for making a Pentium IV chipset was because VIA didn't go to Intel to get a license for the technology. VIA claimed that when they acquired S3 they also acquired the license to utilize the Pentium IV bus technology (since S3 had a license), and it's been fought out in the courts since then.
Regarding Intel's hostility to third-party chipset makers, that only makes sense. After all, making chipsets for their CPU's is a large portion of Intel's business. By licensing their bus protocols to third parties Intel is making sure that they get a cut of every Pentium 4 chipset sold. They're also raising the costs of competitors chipsets to put them roughly in line with their own. Given the choice in that situation, most people would go with Intel.
Also keep in mind that controlling the chipsets also allows you to control the technology that is used in them. The Rambus memory fiasco is an excellent example of that. Rambus turned out to be an expensive dud on the early Pentium 4 systems, but Intel was contractually obligated to support only Rambus RDRAM memory and no other memory type on the Pentium 4 for a certain period of time. During that time VIA was producing a less expensive and better performing SDRAM-based chipset for the Pentium 4. Most people went for VIA chipsets on their Pentium 4 systems and that was hurting Intel's chipset business, so Intel of course attacked in any way it could.
...except for one little consideration. NVidia has not released full drivers for Linux for the nForce. Are they going to be any better for nForce 2?
I am definitely looking at an nForce 2 based solution to upgrade a Windozer of mine, but this would be a splendid solution for Linux if they had the drivers for it. I hope NVidia gets on the ball this time.
Knowledge is power. Knowledge shared is power multiplied.
Now, if somebody like Jumptec, Ampro, or any of the other embedded CPU board makers would use this! I'd love to have that for my embedded system - fast graphics for all the traces, USB 2.0 for RF control, two Ethernet ports for access...
I wonder if anyone could pursade nVidia to put one of these in there... They have everything else....
www.eFax.com are spammers
They actually wrote "I noticed that a eval of NVIDIA's...."
Yahoo's code changed it for you.
(Yes, I know "a eval" sounds horrible... so sue me)
Whoever stated that signature sizes should be limited to one hundred and twenty characters can just go ahead and kiss my
"since a GeForce4 MX is a stripped down, cheaper version of the real powerhouse GeForce4 TI, "
.plan about it: http://webdog.org/plans/1/.
As posted many times by many people, the GeForce MX 4 is a GeForce 2 core with a higher clock speed. The only reason it has GeForce and 4 together at all is because the marketters at nVidia knew they could sell more parts.
Read Carmack's
--
Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
the MX series of GeForce 4 GPUs are woefully underpowered, wait for the next revision of the motherboard that has a non MX GPU. This is such a scam and so many people are getting burned because they see a cheap Geforce4 card and think it is a great deal, when it is just a fast GeForce 2, spread the word!
For the car, I use my portable jukebox, which is a toshiba libretto, mounted using a radio-shack cd player holder (damping and everything). It cost me $390 on ebay for the computer (libretto 110CT), and $130 at a computer show for a 20 GB drive for it. It's a lot easier to just plug that thing into the speakers in whatever room I want to listen in (in the past I had a main server and just had speakers wired to it all over the house, but still needed another computer to control it anyway)
This is incorrect. The chipset includes a TV Encoder, i.e. supports "TV Out" - S-Video or composite out to a TV. From the press release:
NVIDIA nForce2 Platform Processors offer a staggering array of features including:
* TV-encoder and HDTV processor for optimal visual quality
It does not include a TV Tuner capable of receiving broadcast TV. You'll have to add one yourself.
BTW, if you're wondering, the HDTV processor simply means it is capable of decoding HDTV-format MPEG2 video. You would still need an HDTV tuner/receiver to get the signal first.
Why would anyone engrave "Elbereth"?
Since Nvidia doesn't have a license to develop for the intel bus, this will interface to AMD processors (uh, despite that the xbox is intel-based). A version for the Hammer is "far along" and may merge north and south bridge functions into one chip.
Four Taiwanese motherboard manufacturers, including Asus and Chaintech, will use the chips
A future version for server line cards may include gigabit ethernet, routing capability, and a HyperTransport link to network processors.
HIV Crosses Species Barrier... into Muppets
But why throw out the entire PC when all you need to replace is the graphics card? Or sound card or even processor? Surely one of the key things about having modular PC's on a tight budget(or even a reasonably generous one at that) is that you can do incremental upgrades to various system components - which although being more expensive in the longer term, allows you to see the benefits of upgrading in the shorter term.
What are you refering to here? The costs related to just upgrading the video card to upgrading a console?
Let's see, upgrading from one console to the other "next-big-hot-thing" costs about $200-$300. Upgrading from one video card to the other "next-big-hot-thing" costs about $300-$400. Where's the cost savings?
And you dont necessarily have to throw away your old GFX card. Used PC components and consoles have a pretty good used parts market. I let many of my old components go that way if I dont distribute them in my other machines. Though I have hung on to my old consoles.
Your video card is COMPLETELY useless without another computer to plug into. The XBox isn't useless just because you get a new console. The XBox doesn't have to be thrown out when you get an XBox2, just like I didn't throw out my N64 when I got the gamecube. Oh yea, and the used Console market is very very very good (better than the used pc components market.)
So basically, you have no argument.
I'm still waiting for Linux drivers that can legally be linked into the kernel. nVidia forces me infringe on the GPL when I use their nvnet driver. It would also be nice to have the nforce drivers in the kernel distribution.
-Mike_L
The question is how many of those ports are shared.
Usually on a standard 2 port setup, each port gets 12mb or 480mb bandwidth.
If you plug a hub to first port, then all the ports on that hub share the bandwidth of that one motherboard port.. [so on a 2 port system, you can have one badly behaving device hog up all the bandwidth on that port; leaving little for other devices]..
If all 6 ports have individual controllers, then this WOULD be a good thing... and the more ports the easier it would be to segment your USB devices especially if you mix 1.1 and 2.0 devices... It also allows you to group your devices/hubs in such a way that sharing "Groups" between machines is easier.. [unplug groupA from main PC to connect to laptop] or [stuff connected to groupB not supported in Linux yet, so unplug port3 when booting Linux]
btw: I am a "USB power user" with close to 14 USB devices connected at once....
--
Time is on my side
What the guy is saying is that the premium you pay for the system being upgradable is much greater than the cost of considering the computer to be upgradeable.
Two years ago I spent $5K on a home PC. Today the only part of the system that is not obsolete is the video card which I replaced about 6 months ago because the old one was not supported by Windows XP (Vendor went out of business)
The nVidia chipset being discussed is targetted at what are likely to be low end machines, but with much better graphics than are ususal for that sector. OK so the graphics are not cutting edge for gamers, but they are probably better than the cards that ship with 90% of PCs today.
The real advantage to me of a system like that is that you could have a complete system in a slimline case without clunky riser cards, something that I might want to have in the living room next to my HiFi.
Now such a system might not be what I would buy for myself as a primary system but I am certain that my 17 month old son is not going to complain about the lack of performance.
If all you need to add is a processor, a case, memory and some disk you could have a complete system for $500 that you could trust to run unattended without fearing it would burn the house down.
Looking for an Information Security student project suggestion?
Try http://dotcrimeManifesto.com/
Thanks for clarification. It may help that unfortunate Linux user I was replying to.
Count this as a flame-bait, but this IS the Linux' weak point. Once installed, you need to scramble around for tools like this.
And this is not a general purpose software, which is, understandably, maintained outside of a particular OS, but a Linux-specific tool, that, for some reason, is not available with the Linux itself.
No, I like my FreeBSD -- which is an entire OS, not just a kernel, where the existing kernel features quickly get the user-land hooks to use them... Yada-yada-yada :-)
In Soviet Washington the swamp drains you.
Apparently I wasn't the only one who wished that Apple would team up with Nvidia.
If there isn't an expandable Mac around $900 by the time the Shuttle SN40 arrives, I'm getting an nForce2.