[dojoboy]

Looking to grab a new iMac, is the GT 130 worth the bump?

[PeopleLikeFrank]

I couple of things I have read indicate that the 120 is a rebrand of the 9500GT while the 130 is a rebrand of the 9600GSO. This may not be 100% accurate, but the performance difference isn't insignificant: some PC benchmarks. Mac benchmarks show little difference in Q4, a bigger one in CoD4. The PC benches obviously aren't totally apple-to-Apple (har), but they at least have a slightly expanded set of games.

Worth $400? Not sure, but the CPU and HD bump together with the faster GPU might be.

[dojoboy]

Thanks for the info.

[Sneaky Snake]

dojoboy, on May 20th 2009, 11:51 AM, said:

Thanks for the info.

the 4850 is only $50 more then the GT130 and is definitely a big performance boost

[J'nathus]

nobody, on May 20th 2009, 07:57 AM, said:

some PC benchmarks. Mac benchmarks

I'm going to vent a little right here . . . right now.

Why the H E DOUBLE HOCKEY STICKS do they review lower end / mid-range video cards on a system with a top-o-the-line CPU.... How many people in the market for that sort of card is going to have that kind of CPU . . .

End venting . . .

Carry on.

[Sneaky Snake]

J, on May 20th 2009, 08:41 PM, said:

I'm going to vent a little right here . . . right now.

Why the H E DOUBLE HOCKEY STICKS do they review lower end / mid-range video cards on a system with a top-o-the-line CPU.... How many people in the market for that sort of card is going to have that kind of CPU . . .

End venting . . .

Carry on.

they do that to ensure that the GPU is not CPU limited in any way

[J'nathus]

Sneaky Snake, on May 20th 2009, 08:57 PM, said:

they do that to ensure that the GPU is not CPU limited in any way

I know . . . I KNOW! . . . but it still seems AWFULLY silly. If it is CPU limited, then they should state the evidence to that effect. I mean, why don't they get the most expensive i7? Why not a dual processor (not dual core) machine??? I think the review should be more representative of the machine that GPU would find itself in. *sigh* I know . . I KNOW.. If I don't like it, why don't I start up a hardware review site ? ? ? Well because THEN I would be too busy to spend time criticizing how others do things.. THAT'S why!

[teflon]

because this way everything is comparable, from the bottom of the pack to the top, you can go to these reviews of GPUs and see the real difference between the GPUs.

I mean, you might as well say that they should find the best settings to play the game at instead of always picking 1920x1200 and "high" or "Maximum" settings.

Basically, if they do it your way then it doesnt even resemble a scientific test, and therefore the results are invalidated.

[J'nathus]

teflon, on May 21st 2009, 12:46 AM, said:

because this way everything is comparable, from the bottom of the pack to the top, you can go to these reviews of GPUs and see the real difference between the GPUs.

I mean, you might as well say that they should find the best settings to play the game at instead of always picking 1920x1200 and "high" or "Maximum" settings.

Basically, if they do it your way then it doesnt even resemble a scientific test, and therefore the results are invalidated.

So long as the test system is the same outside of the video card, then the base line is there. There are review sites that follow this methodology, so my theory isn't without merit. And in this particular comparison, it isn't bottom of the pack to the top... it's just a handful of mid range GPUs. Now if you were talking about the GPU charts found at Tom's Hardware where you can compare some pretty low end GPUs from the 6 series and up all the way to high end triple SLI and Dual Crossfire systems of today . . . THERE I would agree with your appraisal.

[teflon]

Generally, I believe that a single generation of cards will be put into quite a similar machine, so within a generation itll be quite easy to compare between the cards.
Also, putting the cards into a fairly high end machine as opposed to one modelled on an iMac or something means that the test is solely focussed on the GPU, even if its taken slightly out of context, as opposed to essentially becoming a test of that machine.

I see where youre coming from, I just dont really think it needs to be changed. The current system puts each card into the context of at least is counterparts, and from there you can see if its worth spending the extra $50 for the faster one or if you should wait for the next generation's cards to hit.

[Sneaky Snake]

If your CPU is remotely recent it should be able to handle pretty much any card with a good Overclock. And speed isn't everything, the CPU in my homebuild only has a 1 MB cache which is pretty noticable when I compare to an e8400 (both chips running at 3 GHz)

[J'nathus]

teflon, on May 21st 2009, 11:50 AM, said:

Generally, I believe that a single generation of cards will be put into quite a similar machine, so within a generation itll be quite easy to compare between the cards.
Also, putting the cards into a fairly high end machine as opposed to one modelled on an iMac or something means that the test is solely focussed on the GPU, even if its taken slightly out of context, as opposed to essentially becoming a test of that machine.

I see where you're coming from, I just don't really think it needs to be changed. The current system puts each card into the context of at least is counterparts, and from there you can see if its worth spending the extra $50 for the faster one or if you should wait for the next generation's cards to hit.

I think you're wrong. Not just a little. A LOT. I wouldn't argue this point otherwise. It's just wrong to gauge performance on mid-range cards on a system with a CPU that will NEVER be mated with those parts in the real world. I can see your argument if the part was a higher end Core 2 or Core 2 Quad costing as MUCH as $200-300 . .. but a $900 CPU . . . Never. As a matter of fact, the sales of that CPU in particular were so astronomically small it's ludicrous to even consider using it in a comparison such as this. People with middle of the road machines will look at that, make a buying decision based on it, and wonder why their performance is nowhere near what those charts reflect. That's because most people have systems that are far closer to the 2 GHz range than the 3 GHz range (assuming Core 2 parts internally), which is only addressing clock speed.

In that review, that system is capable of 40 fps in Crysis at 1680x1050. I run Crysis at mostly medium with only 3 settings set to "high" and I can barely do more than 40 fps in 1280x800 (although I do run it vsync'd . . . I can't stand screen tearing). My video card (according to Tom's Hardware charts) beats the pants off of a 4770, to say nothing for a 4670 . . . The advantage must be the Extreme Edition CPU, so . . to me . . that review / evaluation is pretty much worthless with numbers that are meaningless to me... someone running a better than average rig. And I'll tell you, those cards are NOT on my shopping list as they would all be a bit of downgrade from my 8800 GT in all realms except perhaps power consumption.

[teflon]

Alright, Ill concede that it would be quite nice to have all GPUs tested in numerous configurations, low end, mid range and high end. But if I had to choose only one, Id pick a high end CPU in the test every time in order to show the GPUs maximum potential.
They might be worthless to you, but with enough hunting and comparisons (since you should never trust a single review of a GPU or CPU anyway because they are so peculiar to a given setup, site, tester and interpretation of the results) you can pick out the right card.

[Sneaky Snake]

J, on May 21st 2009, 04:04 PM, said:

My video card (according to Tom's Hardware charts) beats the pants off of a 4770, to say nothing for a 4670 . . .

Having owned a 4670 and 8800 GT, and used a 4770 for over a week of gaming; I can easily say that in general the 4770 will be a better buy then a 8800/9800 GT. In general it has better performance, extremely low power requirements thanks to 40nm (meaning that I can Xfire two of them on my Corsair vx 450 watt PSU), incredible xfire performance (performance is in the GTX 275, 4890 range), and real cheap. In my personal tests the 8800 GT had the advantage of being in the Mac Pro with Dual Qaud 2.8 GHz CPU's powering it in Crysis. The 4670 and 4770 only had a 3 GHz dual core with one MB of L2 cache. Despite all that the 4770 came out on top.

Here's a review page compared the two cards in dozens of confirgerations

-Snake

[J'nathus]

Sneaky Snake, on May 21st 2009, 09:02 PM, said:

Having owned a 4670 and 8800 GT, and used a 4770 for over a week of gaming; I can easily say that in general the 4770 will be a better buy then a 8800/9800 GT. In general it has better performance, extremely low power requirements thanks to 40nm (meaning that I can Xfire two of them on my Corsair vx 450 watt PSU), incredible xfire performance (performance is in the GTX 275, 4890 range), and real cheap. In my personal tests the 8800 GT had the advantage of being in the Mac Pro with Dual Qaud 2.8 GHz CPU's powering it in Crysis. The 4670 and 4770 only had a 3 GHz dual core with one MB of L2 cache. Despite all that the 4770 came out on top.

Here's a review page compared the two cards in dozens of confirgerations

-Snake

I tried upgrading to the GTX 260, but it was actually only about 5% faster in some tests and actually slower in others. Possibly my system is holding it back, which is why I haven't pursued an upgrade much further. I have a Quad Core Q6600 (stock clock), 4 GB of Corsair low latency RAM, a SoundBlaster X-Fi Titanium Pro and a 8800 GT with 512 MB of RAM on a motherboard with an Intel P35 chipset. I hope it isn't the motherboard that's holding me back, but if it is... that's a hellish part to upgrade.

As far as power consumption goes, I'm not unhappy with where I'm at, and assuming those charts on Tom's hardware are correct, I'd gain nothing changing to a 4770 and only the tiniest fraction going to a 4850 or a 9800 GTX+. I think my system will have to stay intact / un-upgraded until I finally get the cash for a full system build again.

In the end, it's all down to your 'killer app' or the game you find important, and the latest Radeons don't seem to like Mass Effect (one of my favorite games) . . . that and I prefer the nVidia profiles that allow me to easily force AA in one game, but not the entire list. I am not against ATI . . . as a matter of fact, I'm sort of against nVidia and the crap they've pulled lately, but when it comes down to dollars, sense, performance and controls... nvidia has what I want.

[Sneaky Snake]

There's definitely a problem in there if the 260 only gave a 5% increase. Providing your mobo can handle it I'd overclock your CPU to at least 3 GHz. What mobo do you have?

[dojoboy]

Sneaky Snake, on May 20th 2009, 05:21 PM, said:

the 4850 is only $50 more then the GT130 and is definitely a big performance boost

Yes, this is what I will do.

[J'nathus]

Sneaky Snake, on May 22nd 2009, 08:39 AM, said:

There's definitely a problem in there if the 260 only gave a 5% increase. Providing your mobo can handle it I'd overclock your CPU to at least 3 GHz. What mobo do you have?

It's an ASUS P5K Deluxe WIFI. The fastest motherboard of the P35 release frame that I bought it in, before it was superseded by the next fastest Intel chipset.

In that realm, I have 2 potential issues. The P5K is PCi-E 1.0 and the GTX is a 2.0 card. THAT might be the bottleneck, but I'm not sure. I've thought about overclocking my CPU, but this is a production machine that I do a lot of work on and I can't afford for it to suddenly go down.

I was fearful that my PSU was the problem, but the rating on the card says you need 38A on the +12V rails, well I have 4 rails at 16A (16x4 = 64!?!?!?), which is my understanding as to how these things are figured out. This is also supported by my reading on AnandTech that the GTX 260 as well as the 9800 GTX+ both don't pull more than 9-10A at load. I used 1 PCI-connector for the GTX 260 and 1 y-cable connecting 2 dedicated +12V rails to the card, so it had 16Ax3... that should've been enough.

SO . . . the bottleneck is either my processor or the PCI-E slot. I tried the latest video driver in both an existing install of Vista and Windows 7 RC, and the performance was the same in both, suggesting that there's not a software cause for the issue.

[Sneaky Snake]

J, on May 22nd 2009, 05:46 PM, said:

SO . . . the bottleneck is either my processor or the PCI-E slot. I tried the latest video driver in both an existing install of Vista and Windows 7 RC, and the performance was the same in both, suggesting that there's not a software cause for the issue.

the slots are 16x times though right?

[J'nathus]

Sneaky Snake, on May 22nd 2009, 08:07 PM, said:

the slots are 16x times though right?

Yes, but not if I use it for crossfire (something I never intended to do).