Geforce 4200 beating 4600?!?!

[Byte]

Anyone know why this happens? According to Firingsquad and HardOCP the 4200 trumps the 4600 in a few benchies.

 

[John]

I was asking myself the same question last night.

 

[Killrose]

It will be interesting to actually see the actuall 4200 spec when it is released. I have a feeling there is going to be a special bios like the Radeon Retail vs LELE to hamper performance.

If not, Nvidia will be shooting themselves in the foot. Which I beleive they have already done.

 

[FuManStan]

<< Anyone know why this happens? According to Firingsquad and HardOCP the 4200 trumps the 4600 in a few benchies. >>



Trumps? The only victories for the 4200 were in 3dmark's high detail car chase by 1 frame, and in Serious Sam 2 by .1 frames in Firingsquad's review. Margins that small are negligable and can be attributed to other parts of the hardware.

 

[John]

<< Margins that small are negligable and can be attributed to other parts of the hardware. >>



Yes the margins are small, however a lower clocked card is beating the "flagship" model. Both were tested on the same machine.

 

[m1ke101]

wasn't the card overclocked? or am i mistaken.

 

[Emo]

No, it wasn't overclocked. They were comparing two versions of the 4200, one with 64MB DDR running at 250/500, and the other with 128MB running at 250/444. In some benchmarks the Ti 4x00 were all very close but the 4200 does not "trump" the 4600 or even the 4400. Nevertheless, I think the 64MB 4200 card is the best value since for half the price of the 4600 it delivers 85-95% of the performance.

 

[FuManStan]

<<

<< Margins that small are negligable and can be attributed to other parts of the hardware. >>



Yes the margins are small, however a lower clocked card is beating the "flagship" model. Both were tested on the same machine. >>



The point is that the margins are too small to be attributed to the speed of the card, meaning the benchmarks in question are CPU limited and have nothing to do with the difference in video card.

 

[PUNKtotalled]

Conclusion: I want a Ti4200.

 

[BFG10K]

Anyone know why this happens? According to Firingsquad and HardOCP the 4200 trumps the 4600 in a few benchies.

The Ti4200 is the exact same card that the Ti4600 is, just clocked lower, hence there is no possible way that it could have beaten the Ti4600 by a significant amount. Either the Ti4200 was overclocked, the benchmarks were flawed or there was something wrong with the Ti4600.

As for small victories (eg 1-2 FPS) it'll simply be benchmarking noise, especially if the tests were run in CPU limited situations.

 

[PUNKtotalled]

<< Anyone know why this happens? According to Firingsquad and HardOCP the 4200 trumps the 4600 in a few benchies.

The Ti4200 is the exact same card that the Ti4600 is, just clocked lower, hence there is no possible way that it could have beaten the Ti4600 by a significant amount. Either the Ti4200 was overclocked, the benchmarks were flawed or there was something wrong with the Ti4600.

As for small victories (eg 1-2 FPS) it'll simply be benchmarking noise, especially if the tests were run in CPU limited situations. >>





....or it is a driver issue.

 

[Gonad the Barbarian]

<< Anyone know why this happens? According to Firingsquad and HardOCP the 4200 trumps the 4600 in a few benchies.

The Ti4200 is the exact same card that the Ti4600 is, just clocked lower, hence there is no possible way that it could have beaten the Ti4600 by a significant amount. Either the Ti4200 was overclocked, the benchmarks were flawed or there was something wrong with the Ti4600.

As for small victories (eg 1-2 FPS) it'll simply be benchmarking noise, especially if the tests were run in CPU limited situations. >>



The 4400 & 4600 are using BGA RAM, the 4200 is not. Maybe the latency is higher on the BGA RAM even though it is clocked higher.

 

[BFG10K]

....or it is a driver issue.

With the same core it won't be.

Maybe the latency is higher on the BGA RAM even though it is clocked higher.

It's certainly not going to be so high that the massive clock speed difference is neutralised.