• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Theory - The highest clocked GPUs in a series tend to be the best value

g67575

g67575

g67575

Soldato
Joined
30 Jun 2019
Posts
8,159
So, cards like the RX 6600, RX 6700 XT, RTX 3070 and RTX 3070 TI. They all have the highest clock rates in their respective series line ups, which you can see here:
https://en.wikipedia.org/wiki/GeForce_30_series#GeForce_30_(30xx)_series
https://en.wikipedia.org/wiki/Radeon_RX_6000_series#RX_6800_and_6800_XT

I could add in the RTX 3080 too, but it's clocks are slightly lower than the RTX 3070 and TI. There's little difference, so the effect isn't noticeable here.

The higher end you go though (above the RTX 3080), the less substantial the performance gains are.

An exception to this rule is the RTX 3090 TI, which will be a very expensive card, low yield (that is likely a difficult product to produce) with a boost clock upto 1860. It's an outlier and quite late to the party.

I think the fact the highest spec cards often have lower base and boost clock rates indicate a hardware / technological constraint, perhaps related to the current 8nm and 7nm manufacturing processes in use.

If that's true, you can expect diminishing returns at the higher end, as these cards have difficulty maintaining stable clocks across a greater number of GPU processing units.

Perhaps with 5nm and 6nm GPUs, this rule will apply to a much lesser extent? As power usage should be lower, and higher clock rates are often possible with new fabrication technologies. One example of this, is AMD predicting 5ghz all core clocks on CPUs built with 5nm fabrication technology.
 
Last edited:
ConfusedMistyBalloonfish-max-1mb.gif
 
Tired9 said:
6500 xt says no.
Click to expand...
Actually, this kind of proves my point, with official pricing at low as £200 including VAT. That's less than a series S console.

Not saying it will be available at prices anywhere near that, especially since no one will be able to buy at RRP anyway, due to no reference model.
 
g67575 said:
Actually, this kind of proves my point, with official pricing at low as £200 including VAT. That's less than a series S console.

Not saying it will be available at prices anywhere near that, especially since no one will be able to buy at RRP anyway, due to no reference model.
Click to expand...
A series S is substantially faster, order of magnitudes faster.... The XBOX ONE X is seriously far faster too LOL and that's a custom Polaris chip in that.

https://www.techpowerup.com/gpu-specs/xbox-one-x-gpu.c2977
 
That's almost certainly incorrect, check the specs here:
https://www.techpowerup.com/gpu-specs/xbox-series-s-gpu.c3683
https://www.techpowerup.com/gpu-specs/radeon-rx-6500-xt.c3850

The 6500XT has a higher pixel rate, texture rate and TFlop count.

It's weakness is purely the 4GB of VRAM.

AMD is claiming effective memory bandwidth of the RX 6500 XT is 232 GB/s, theoretically just a tiny bit faster than the Series S, due to it's infinity cache.

Full spec here:
is https://www.amd.com/en/products/graphics/amd-radeon-rx-6500-xt
 
Last edited:
Project Scorpio out paces the Series S console in raw resolution as the S is locked to 1440P max, I can run Forza Horizon 5 at 4K all day long on my One X with zero issues with a locked 30 FPS on high settings, you can't do that with a 6500 XT lol.

Polaris. 2560 shader units, 160 TMU's, 384-Bit and GDDR5.
 
Simple answer, there's less driver /software overhead on consoles. I'm still doubtful there's a noticeable difference vs the Series S.

Yes the Series X is a significantly more powerful GPU (12 TFlops) kind of a tangent though lol.
 
@OP - cherry picking a couple of cards from the weird covid induced situation we are in doesn't prove your theory:

Plenty of other examples that debunk your theory:


AMD
RX590 - highest clock rate of the RX5xx generation - yet the RX570 was better value. Same with the RX580 before the 590 was introduced - the 570 was better value

R7 260X - was highest clock rate of that generation, but 270X offered lots more performance due to wider memory bus and more shaders, so was better value. 280X offered even more shaders, even wider bus, more VRAM and overclockability - so again better value

7xxx series - 7770/7870/7970Ghz Editions all offered the highest clock speeds, but 7850 was probably sweet spot in performance/price.

4850 was better value than 4870 despite lower clock speeds ....

9500 Pro and 9600 Pro were better value than 9700 Pro and 9800 Pro respectively

NVIDIA
Geforce 960 had the highest clock speed, but 970 was the best value card of that generation

Geforce 8600GTS - highest clocks, but the 8800GTS and similar "cut down" big cards would have been better value

Geforce 4 Ti 4600 and Geforce 4 MX460 both had 300Mhz clocks, but Ti 4200 was by far the better value.

Geforce 3 Ti500 had the highest clocks, but Ti200 was the best value card

Geforce 2 Ultra had the highest clocks, but the GTS or Pro were much better value
 
Value is also subjective...

So... like good luck!

I prefer a console at such low end, you know it will run properly with decent settings and will be patched for that specific hardware.

PC is brute force.
 
I suppose you have a different understanding of debunking an idea, than I do.

Especially since I was only discussing modern graphics cards.
 
So, I'm just talking about new cards available to buy now. Many of the cards I mentioned, with the highest clocks have the best value, at 4K resolution. See for yourself:
https://tpucdn.com/review/sapphire-...e/images/performance-per-dollar_3840-2160.png

And here:
https://www.techpowerup.com/review/sapphire-radeon-rx-6500-xt-pulse/31.html

I think the main caveat for us in the UK, is that reference prices for AMD cards aren't an option. But it's certainly true for Nvidia reference models, the highest clocked models have the lowest cost per frame.

One outlier here is the RTX 3070 TI, which spec wise, is just to similar to the RTX 3070.

In the US though, the 6nm RX 6500 XT, with a boost clock of 2815mhz has the lowest cost per frame in all resolutions, at least at the RRP price.

This card also seems to be available in the UK for around £225 new (non reference models).

The older cards in the chart are arguably somewhat irrelevant when looking at cost per frame, as you can't buy them new any more...

On possible prediction you can make from this, is that if AMD releases more 6nm GPUs (presumably clocked significantly higher), these should offer improved value vs current 7nm cards. I've no idea if this would only apply to low end / mid range GPUs, or high end 6nm GPUs also (they may not make higher end 6nm refresh models).
 
Last edited:
Considering a used RX 580 once the market settles will be around 140 GBP max, the 6500 XT loses all arguments it had in it's favor.
Nobody is locked out of the used market to create a control group just so a card can appear better for the money...
 
Well, it has just 4Gb of VRAM, quite a drawback, but not really if playing at 1080p resolution.

I'm not really making an argument in favour of the RX 6500 XT, just pointing out that 6nm appears to be quite a positive development if it allows improved clock rates on RDNA2 refreshes, presumably at a similar cost to current models.
 
You must log in or register to reply here.
Back
Top Bottom