Review NaN of 4
, from London, UKSummary:
Defying the Laws of Nature by conceiving a new graphics card every six months for so long has finally taken its toll on dear old NVIDIA. Alongside the hugely impressive gf4/4600, the masters of visual silicone have produced a dead duck - the gf4MX.. One could conspicuously tell that these two sisters, by name, were separated at birth..
**The Devil’s Spawn**
It has been a bumpy roadmap that NVIDIA embarked on and, perhaps to many, the leap backwards to the days of the gf2MX made little sense..
From a history as prestigious as that of NVIDIA, ‘backwards’ is not a word commonly associated with the inventors of the GPU. The advent of the transformation and lighting (or the stupendous T’n’L Engine), that was successfully implemented in the original GeForce 256 from NVIDIA, led directly to the evolution of the Graphical Processing Unit (GPU). The need to upgrade the CPU in order to help calculations in hardware receded, especially when the gf256 ripped the competition apart!
The thrust to the throne of video card kings did not go about as a subtle war of words, rather the ruthless massacre of existing opponents won what was a frightful sight. The former kings 3dfx simply did not abdicate, instead NVIDIA’s coup de gras was to ensure 3dfx were hung, drawn and quartered – without remorse or pity. What was left of 3dfx merely coalesced into a manageable pulp for NVIDIA to consume and digest.
In a sorrowful end, 3dfx's time was up.. They didn't evolve.. NVIDIA on the other hand began something special - raw speed wasn't everything. What I’ve noticed in the past few years is NVIDIA’s attempts to produce 3D cards that could also compete in the OEM, business, budget and every other sector. I’ll point out the multi-monitor support, the improved 2d application, dedication to a FSAA solution that'll work so well that 3dfx's grave would turn yellow with urine. And don't forget the driver support..
So to commemorate the ‘unmatchable innovation and technological feats, I have decided to follow in the footsteps of NVIDIA, by going back to unfortunate basics..
**A Is For..**
Accelerated Graphics Port, you didn’t really think I had ‘dumb down’ that much did you? The port was specifically intended for high-speed interfacing between the processor and the video card. Older ISA video cards became serious bottlenecks when users moved to graphically orientated operating systems like windows, whilst the bandwidth on PCI video cards is strained by the heavily demanding video performances of 3D acceleration. The birth of the AGP addressed this problem by defining a new interface for video information that quadrupled (or even more) the theoretical bandwidth of current PCI buses.
PCI is an acronym for Peripheral Component Interconnect, introduced by La Intella and assisted in their ever deepening conspiracy with Bi££ Gat€$ µi¢r0$0ft.. As an overview, the PCI based systems made popular the conceptual use of the now taken for granted Plug and Play. The PCI chipset circuitry handled the identification of cards and works with the Operating System and BIOS to automatically set resource allocations for compatible peripheral cards, but the video performances were sacrificed - for a price.
I mention these ‘buses’ (loosely translated as such) because this new card is available in either format.. At least there is some good news.
**Spoilt For Scraps**
When it comes to PCI based cards your options are limited to the voodoo 5500, gf2MX and now the gf4MX.. I must add the voodoo isn't supported anymore, so purchase at your own risk.
Again, NVIDIA spends time on tweaking their products, so the original gf2MX branched to 400 or 200 derivatives ( I lean toward purchasing the gf2MX 400 instead of the 200).. The gf2MX 200 has half the memory bandwidth compared to the original gf2MX card - but then anything is better than an integrated graphics card. If you need some 3d acceleration, at least for the modest games out there today, the gf2MX/400 can arguably suffice on the most stricken of budgets. In any case the MX400 is dirt cheap today, whilst the relatively new gf4MX still has a niche to fill. It’s enough to make your head spin..
From experience with NVIDIA cards: look for memory bandwidth, i.e. the bigger the better (over clocking the clock speed didn't matter a jot)
Gf2MX/400 200MHz core speed; 64mb sdram (166 MHz fsb); memory bandwidth 2.65 GB/s
Gf4MX/420 250MHz core speed; 64mb sdram (166 MHz fsb); memory bandwidth 2.65 GB/s
Neither you nor I am Einstein, but you don't need to be a genius to figure out that there is no real significant difference - anyone pointing to a 50 MHz increase in clock speed has definitely not tried over clocking the core of a gf2MX ;)
So what are the advantages of getting the gf4MX? Better anti-aliasing (but for the price of butter, do you think the gf4MX can pull off something that the far superior gf3ti200 struggled with?). It also has decent hardware support for DVD playback. The gf4MX also features the nview engine, which significantly improves on the much criticised dual display technology of the gf2MX. But best of all: multi-display support, i.e. 2 monitors - you've got to try it to like it! My opinion is to get the gf2MX/400 - it’s perfect for a very strict budget.
**Naughty Name Games**
The gf4MX is available as 460, 440 or 420 (God forgive you if you bought the 420 – it runs almost on empty with a paltry 64md SDR sdram)..
Gf4MX 460: 300 MHz GPU clock and a 275 MHz memory clock .. 64mb DDR (300/500)
Gf4MX 440: 270 MHz with a 200 MHz memory clock.. 64mb DDR (270/400)
Gf4MX 420: 250 MHz core with 166 MHz memory clock.. 64mb SDR sdram (a pathetic 250/166 avoid)
By comparisons sake, ATI Radon 32mb DDR: clocked at 166/166 MHz or 166/333 MHz effective due to DDR
Good news: the card should smack any game running at the present time, graphics should be very smooth and frames-per-seconds should melt butter. Your benchmarks would cream the gf2MX and the 2 years older Radeon cards. That’s it.. Twiddle thumbs
Bad news: the gf4MX would basically be a gf2MX with an improved memory controller, multisampling AA unit, and updated video features. In English>>> a heavily crippled gf3, without two pixel pipelines or DirectX 8 compliance..
**Marriage Isn’t A Word..**
..It’s a sentence. And yes, you read it right.. Not DX8 compliant. The damn NVIDIA lot flashed out another episode of pulling wool over the eyes of consumers.. I’d be embarrassed to call this card a gf3MX. Shame, NVIDIA!
Considering the 18 month difference, NVIDIA managed to haul GPU technology backwards (I truly mean it), resulting with cybernetic cries for the ‘d-words’(don’t tell anyone I said dissociate the dizygotic disaster, okay)! This product should have been called nothing but a gf2MX extension. Forget the SSSCA! The Propaganda kings are at it again..
The gf4MX is a fair card, don't get me wrong.. But it’s not great, the product is a misnomer - what I would recommend: return the card and fetch a gf3 ti200 or ATI 8500le or the true gf4 (4200, 4400 or the 4600 - the latter being the new flagship and with it comes a huge price). The negligible difference in price between the gf4/4200 and the gf4MX/460, yet the disparate performances, leads to a conclusion so obvious to any layman that I’ll leave you to your imagination!
As an aside, I would still recommend the gf3 ti200 over the weaker gf4MX/420 or even the 440 - if you can fork out for a brand spanking new card then surely the extra investment is worthwhile, unless you need the dual display feature of the MX. N.B. gf3ti200 is DX8 compliant. The gf4MX isn’t, therefore not as 'future proof' (bad word.. can't find a more suitable one). Alternatively, convert as I have to ATIism..
But even funnier, the gf4MX can't even cut crud.. Less than 25fps on the new unreal engine at 1280*1024*32! Pathetic.
**Clutching At Amateurish Names**
It’s bad enough that NVIDIA committed the cardinal sin of misleading the public with their awfully designed nomenclature, but for Gainward to repeat the mistake and then compound stupidity beyond belief?
Gainward called their gf4 ‘Ultra’, whilst this gf4MX is labelled as ‘Pro’.. Where shall I start? ‘Ultra’ and ‘Pro’ were tags carried over from the clocked gf2 GTS of old, skipping the gf3, and plastering this sickening piece of graffiti on the gf4 package.. There is no other relationship! Many hardware buffs pronounce the gf2 Pro as the video card with the most bangs for your buck some 12 months ago, so to acknowledge the gf4MX with such an honour is ultimately lazy and ignorant.
And the disaster doesn’t stop there! At this point I have to admit that I do have (ahem) contacts that can provide me (rarely) to run their goods, albeit with a lot of grovelling and shelving any sort of honour. So as I write the reviews for the Gainward gf4/4600 and gf4MX solutions, spare a thought for me.. Please..?
Gainward traditionally displayed the name of the GPU alongside the cards they sold, the gf2 and gf3 showed prime examples of this. However, this logical – and non-confusing – system abruptly disappears with the recent line of products. I opened the ‘Ultra’ and was greeted with a welcoming sign: ‘GeForce4 PowerPack!’. Huzzah!
With, understood by many, less enthusiasm I ripped through the MX packaging, only to be stumped by this sign: ‘GeForce4 PowerPack!’.. Oh, you mean I forgot the tiny detail of the ‘Pro’ tag.. So that makes it all right then, does it?
The baffling episode continues.. The numbers following the Pro/Ultra have no correlation to the shipping clock speeds of the cards themselves. For example, the 4600 card is called the Ultra/750XP but the number stands as virtually redundant. Unless the influence of IBM hard drives spread further than first imagined.. This Intel/ Microsoft/ NVIDIA conspiracy is more elaborate than I first feared..
The messy naming system is certainly not chivalrous, nor does it inspire. Did someone say ‘pigs’..?
**Conclusive Bushido?**
Remember how NVIDIA managed to pull off a dirty campaign to undermine ATI’s r8500 video card, by releasing variants of the gf3 core and the detonator 4 (also known as XP) drivers? Well, if sources of Hexus.com are believed, NVIDIA have plans to release drivers that could ‘offer the same kind of performance boost for GeForce 4 Ti's that we saw with Detonator 4 and the GeForce 3 and steal some thunder from the new announcements from Matrox and the rest (e.g. PowerVR).
’Rumour has it that the GeForce 4 Ti has an unused texture unit going spare on the silicon which the new drivers will mysteriously enable. It's certainly a reason why aniso filtering performance isn't that hot on NV25 from what we can tell and it's explains the slight feeling that NVIDIA are keeping something from us, like they did with GeForce 3. Look out for new drivers soon from NVIDIA, and in any case, they are running out of numbers for the current Det's! 29.xx is awfully close to a new 30.xx release!’
If anything, NVIDIA knows the strengths and weaknesses of their opponents – as well as their own.. Victory is almost secured! Or is it?
I say almost, because this year NVIDIA will need to keep their wits about them.. ATI are no longer the only opposition. Entering the arena are Matrox’s Parhelia-512, 3D Lab’s P10 original VPU (Visual Processing Unit) and to a lesser extent (major in the field of OEM and budget machines) PowerVR’s Kyro 3..
Report this review >>
|
2 of 5
5 of 5
3 of 5
