GPUs & CPUs & Enthusiast hardware: Questions, Discussion and fanboy slap-fights - Nvidia & AMD & Intel - Separe but Equal. Intel rides in the back of the bus.

are you using A1111? it already has a setting for that
View attachment 5817571
adding "--no-hashing" to the launch arguments also helps a lot
I tend to use ComfyUI though I do have Automatic1111. Though looks like I can jerry-rig ComfyUI to do the same. That's not for VRAM, though?

I'll check it out and the no-hashing setting. Cheers.
 
Looked at my finances after paying bills after my paycheck. I'm close but not there to buying a 4070/4070 super. I'm getting tempted at this point to buy a weaker card and get it over with lol.
 
  • Like
Reactions: Brain Problems
FSR is only tolerable on Quality, which rasterizes 45% of the pixels and derives the rest. Even at Quality, it still tends to have visible shimmer and other artifacts.

By contrast, I usually can't tell DLSS is on even in Performance mode, which rasterizes only 25% of the pixels. Frame generation is icing on the cake.
I didn't really care about DLSS before I got a 4070 and it blew me away in e.g. Cold War or Brothers.

I wouldn't be as severe as you are on FSR, but there's no denying DLSS is serious added value. It feels so effortless, just toggle it and your game looks borderline identical except with a massive framerate gain.
 
Intel getting some flak for their "new" i9-14900KS chip. One of the comments made me laugh: "Did they call it the 14900K because that's the operating temperature?" It consumes up to 300W, apparently.

Anyway, apropos to nothing, timestamped some of Hardware Unboxed video on it that gave me a small chuckle. They are not impressed.

 
It's a limited run chip. It doesn't matter, not worth talking about. The people who get it will probably like it. It's more efficient since it's a better binning.
 
It's a limited run chip. It doesn't matter, not worth talking about. The people who get it will probably like it. It's more efficient since it's a better binning.
Ah, didn't realise it was a limited run. I suppose that's markedly less silly on Intel's part though I'd raise my eyebrows at anybody who thought it was worth getting.
 
  • Informative
Reactions: Overly Serious
Chips like this are marketing stunts. Win some online pissing matches, boost the brand identity down-market, hang on to your OEM relationships.

Brand reputation is difficult to manage because of how dumb most consumers are. One outlying benchmark done a particular way can have significant effects on market perception - I know people who thought AMD EPYC CPUs were "broken" after seeing certain benchmarks where, it turned out, the EPYCs underperformed because MKL didn't take the AVX-2 enabled code path. Fix the code path, and the performance was great. Explaining this to them was like trying to explain chess to a cow. They went out and told their customers to not buy EPYC.
 
Chips like this are marketing stunts. Win some online pissing matches, boost the brand identity down-market, hang on to your OEM relationships.
Intel gets a new wave of reviews that retest with the latest drivers, the benefit of a larger roster of APO games than was out when the 14900K launched, and clawing 1-5% extra performance from clocks making them look better on benchmark charts. AMD did their own stunt with those pointless 3900/3800/3600 XT CPUs.

Speaking of chart toppers, I still hope we see AMD do a 24-core Zen 5 X3D CPU (8 Zen 5 cores with 96 MiB 3D V-Cache, 16 Zen 5C cores with 32 MiB).
 
  • Thunk-Provoking
Reactions: Overly Serious
Well in this case I think the marketing stunt is backfiring because they've pumped enough juice into this thing to power a house and it's still performing behind the 7800X3D from what I saw. When it comes to marketing, the old adage about "better to have tried and failed..." doesn't necessarily hold true.
 
Well in this case I think the marketing stunt is backfiring because they've pumped enough juice into this thing to power a house and it's still performing behind the 7800X3D from what I saw. When it comes to marketing, the old adage about "better to have tried and failed..." doesn't necessarily hold true.

It's ahead of the 7800X3D in some benchmarks, and behind in others. What a YouTuber is going to say will then depend on just how many "ahead" benchmarks they run and how many "behind," because most of them just do a linear average of results. What the benchmarks actually show if you dig in is which software applications are memory-bound, and which are compute-bound.

The 7800X3D is going to have a clear advantage in any memory-bound workload, while the i9-1400KS will have the advantage in any compute-bound workload. If I benchmark 10 memory-bound workloads and 5 compute-bound, I will proclaim 7800X3D the winner, and if I reverse that ratio, i9-1400KS will get the soyface thumbnail.

This is why I very much dislike consumer-oriented marketing. Not too long ago, I was in a situation where consumer marketing fags were trying to infect the enterprise side with their bullshit tactics. It was extremely annoying.
 
It's ahead of the 7800X3D in some benchmarks, and behind in others. What a YouTuber is going to say will then depend on just how many "ahead" benchmarks they run and how many "behind," because most of them just do a linear average of results. What the benchmarks actually show if you dig in is which software applications are memory-bound, and which are compute-bound.

The 7800X3D is going to have a clear advantage in any memory-bound workload, while the i9-1400KS will have the advantage in any compute-bound workload. If I benchmark 10 memory-bound workloads and 5 compute-bound, I will proclaim 7800X3D the winner, and if I reverse that ratio, i9-1400KS will get the soyface thumbnail.

This is why I very much dislike consumer-oriented marketing. Not too long ago, I was in a situation where consumer marketing fags were trying to infect the enterprise side with their bullshit tactics. It was extremely annoying.
What's the difference in cost, availability, and wattage between the two chips?
 
Its weird to me that DLSS/FSR is seen as anything but better anti-aliasing. This is always in some sense going to be just like stretching out a sub resolution signal onto your monitor for sake of frame rate, as we all did in the good old days.

Personally I have always had better luck making out the detailes of an un-stretched image than a badly stretched one, so I assumed there was a lot of room for improvement, but its never actually going to be the same.
 
It's ahead of the 7800X3D in some benchmarks, and behind in others. What a YouTuber is going to say will then depend on just how many "ahead" benchmarks they run and how many "behind," because most of them just do a linear average of results. What the benchmarks actually show if you dig in is which software applications are memory-bound, and which are compute-bound.

The 7800X3D is going to have a clear advantage in any memory-bound workload, while the i9-1400KS will have the advantage in any compute-bound workload. If I benchmark 10 memory-bound workloads and 5 compute-bound, I will proclaim 7800X3D the winner, and if I reverse that ratio, i9-1400KS will get the soyface thumbnail.

This is why I very much dislike consumer-oriented marketing. Not too long ago, I was in a situation where consumer marketing fags were trying to infect the enterprise side with their bullshit tactics. It was extremely annoying.
Fair points but in this case, or rather this market, the context is going to be "games" and my impression is that the 7800X is for the most part still ahead there.
 
Fair points but in this case, or rather this market, the context is going to be "games" and my impression is that the 7800X is for the most part still ahead there.

Some games are memory bound, some aren't. Intel cherry-picked compute-bound games for its marketing slides. YouTubers will pick a different mix of games. The point is you can get any result you want by cherry-picking.

If I were an Intel fanboy reviewer, I could pick a dozen games to make Intel look good, do my soyface thumbnail, put ULTIMATE GAMING CPU on the headline, and get 10,000 clicks.

What's the difference in cost, availability, and wattage between the two chips?

It's a marketing stunt, so none of that matters.
 
Some games are memory bound, some aren't. Intel cherry-picked compute-bound games for its marketing slides. YouTubers will pick a different mix of games. The point is you can get any result you want by cherry-picking.

Yes, you can. But I'm hoping channels like Hardware Unboxed are mostly puling out a fair and representative sample. In any case, the best Intel can say is that it's close. And given the following question and answer:

What's the difference in cost, availability, and wattage between the two chips?
At my favourite Bongland store, it's £689.99 for the i9-14900KS and £349.99 for the 7800X3D. So that's a difference of £340 or, in Burgernotes: $433.

Note that the AMD chip has been out for a while and the Intel one hot off the press so would be more expensive. I'd say it would come down but from what @The Mass Shooter Ron Soye said it's a limited run so maybe it wont!

Anyway, to make the point I alluded to, this is terrible value for money against current competition. And if you are someone who has the in-depth knowledge to find one is better for your specific and probably non-gaming use case than the other, you're probably not buying either of these chips in the first place.

EDIT: An oh yes, what @jeff7989 said about power consumption, too!

It's a marketing stunt, so none of that matters.

Matters? Maybe not. But signifies? Yes - something. Because if this is what Intel can manage as a "marketing stunt" it's pretty bloody pathetic. Each year they re-release the same chip with marginally higher clock speed. I'm not arguing with you that it is a stunt or even if it really matters that much. I am saying that the marketing stunt is backfiring because all it's doing is presenting to the world an Intel chip which is for its market slightly worse than a rival chip which costs half as much and for which they've only achieved this much by throwing MOAR POWAH! at in a blatant attempt to compensate for their inability to improve by other means.
 
Last edited:
None of this shit matters! But here's a chart that Intel provided:

RPL-S 14900KS Slides (5)_575px.png

The fact that 7950X3D is ever significantly less than 7800X3D shows the downside of the messy mixed 3D cache / higher clocks approach, which is why I'm interested in the 24-core I mentioned.

7800X3D wins in 3 of these, and APO is enabled in 2 of the 6 Intel wins. Either way, 7800X3D is a better gaming CPU and very power efficient, with a big hit to multi-core performance for those that care. Oh how the turns tabled.

Matters? Maybe not. But signifies? Yes - something. Because if this is what Intel can manage as a "marketing stunt" it's pretty bloody pathetic. Each year they re-release the same chip with marginally higher clock speed. I'm not arguing with you that it is a stunt or even if it really matters that much. I am saying that the marketing stunt is backfiring because all it's doing is presenting to the world an Intel chip which is for its market slightly worse than a rival chip which costs half as much and for which they've only achieved this much by throwing MOAR POWAH! at in a blatant attempt to compensate for their inability to improve by other means.
In the TechTuber / tech news echo chamber, Intel is derided for this KS launch, who knows how many consoomers pay attention to that. There was a rumor of internal dissent at Intel on whether it should have been released since it makes them look bad and has almost no volume anyway. But in the end I don't think anyone will be so repulsed by this novelty CPU that they will change buying habits.

I hope we see some better stunts from Intel soon, and change is coming. Arrow Lake is tile-based, and it's conceivable that Intel could put Adamantine L4 cache on Arrow Lake desktop chips as their own answer to X3D.
 
  • Agree
Reactions: Post Reply
Back