New Gaming PC Build Questions

was talking about the world's biggest monopoly and the single critical point/bottleneck; ASML
https://www.datacenterknowledge.com/hardware/us-sanctions-push-china-create-own-chip-tech-asml

https://www.bbc.com/news/business-64514573

Agree with you guys on value for money. to be honest the last real great value-for-money leap in GPU's from Nvidia was the 2080ti and 2070. and before that the 1080ti. But its a profit and survivability game.
But pure performance-wise, the 4000 series kick the living *** out of the 3000 series. I upgrade my GPUs every 2nd or 3rd generation.
Up until 2016 I was still playing on my old 690GTX Sli rig. I will skip both the 4000 and 5000 series, and probably get the 6000's sometime in 2026/2027.
the 3090's will be more than enough firepower for anything until then, and while someone with a 13th gen CPU and 4000 series will have 30-50 more FPS on ultra settings on the latest high end AAA games, I'm not bothered as the games I play like diabloIV, POE, FFXI would still run fine on a 1080ti at max settings.
But if i was going to build a new rig today, I would go for the 4000's.

They are a Dutch company aren't they? They sell the extremely expensive and advanced lithography machines, to the tune of a few hundred million each. Them not selling to China just means more inventory for other chip factories since ASML tends to have multi-year backlogs.

People freaking out about the ban on selling tech to China are doing so because they have financial investments into Chinese tech companies. Anyone who has a financial investment into a Chinese tech company, which is most of your elected officials, is likely to take a lose from the sanctions, unless a workaround can be created.

4090 only, the rest are marginal within their price brackets.

That is what makes generation suck so badly, nVidia is using the generational performance gain as an excuse to readjust the product segments. A 4050 became a 4060, a 4060 become a 4070 and so forth, then adjusted prices upwards to match previous crypto mining level price-vs-performance. The only card this wasn't done to was the ridiculously priced Halo product, previously called a Titan then renamed xx90. This is what makes it "Malicious Segmentation", the entire production lineup is priced in such a way to upsell on the 4090, which has the highest profit margin for nVidia.

It's also why I tell people to avoid the 40 series right now, any purchase is a bad purchase. Just wait until after the fall releases to gauge where things are headed, and if possibly just skip this entire generation. The market will settle the issue within a year or two, especially since Intel's Arc is actually putting up good numbers in the midrange.

Yeah. 4080 may be a bit better than a 3090 but the uplift isn't as significant as the increase in price for an x80 card. 4090 was the only card that really felt like a generational performance leap and it has titan pricing.

I remember buying 2 780s when they were new for less than it costs to get a single 4080 today. Insanity.

I am part of the problem, I bought a 4090 and I'll probably buy a 5090. I guess I should stop doing that.

xx90 series are never good price/performance wise. Best bang for buck is the xx70 line. Can replace every 2 years or do for much less than a single xx90.

For every generation before the 40xx yes, the 4070, like the rest is just terrible. Right now I'm recommending anyone who absolutely has to buy a card to get a 6700XT. $300~400 USD for very good mid range numbers.

https://www.amazon.com/XFX-Speedster-SWFT309-Graphics-RX-67XTYJFDV/dp/B08YKCC8XD/ref=sr_1_2?keywords=amd%2Bradeon%2Brx%2B6700%2Bxt&qid=1688567575&sr=8-2&th=1

The best way to look at this stuff is USD/year spent, $400 every two years is $200 per year. $1700 every 3 years is $566 per year spent. Someone would have to keep that 4090 for eight years for it to have the same value as that midrange card, and lets not kid ourselves those folks are going to buy the next Titan product the moment it comes out. This is why I always push people to write down a list of requirements and then do product selection based on those requirements, HW vendors are way to good at convincing people to buy the latest greatest.

It pretty much comes down to, I buy the best card because I have the means. If I were budgeting out a build where I can only spend $x and not a penny more I'd skip 40 series entirely.

was looking at an analysis report on the entire supply chain, and the implications of the fresh restrictions,
tl;dr/highlights
-Dongfang being shady. no controls in place to keep them at bay.
- US recognised urgency now to restrict access to ASML machinary and equipment
- ASML reached agreement with US
- US doubling down on restrictions
- Case being brought forward by China against US within the context of B2B and not G2G, where Chinese private companies attempting to sue US government through investment protection treaties
- in March, Nvidia announced that it is working with ASML, Taiwan Semiconductor (TSMC) and Synopsys (SNPS) to use artificial intelligence (AI) to accelerate semiconductor breakthroughs.22 May 2023
I'm a bull for nvidia and a bear for the sector.
watch this space:

and it's not looking good supply-wise. Definitely a trigger for speculators.
little nuggets in this episode by the way, although it's filled with a lot of hyperbole and typical fear-monguering. but should watch

I agree with all of this. As a consumer its a kick in the teeth. as an investor its brilliant business strategy.
I'm not holding my breath for the intel GPU's as a consumer or an investor for now.

ahahaha
As I read this, after Saev's breakdown of cost for gpu/year, my emotions went from "yeah I haven't approached this rationally. I'm feeling guilty, stupid and need to recognize I have a habit for buying big shiny things"

but then I read the end and felt comfort that there are other people who are just as guilty and smart.

It's understandable, marketing teams are extremely good at manipulating consumer emotions to convince them to buy stuff. They employ a whole host of techniques all aimed at generating a desire to buy their stuff, then helping you rationalize the purchase. The only real defense is to write down solid requirements and adhere to them when making purchasing decisions, especially if it makes you feel bad.

I don't think the market is responding well to nVidia's recent consumer GPU product placement scheme. Consumer GPU sales are in the toilet, sellers are already starting to discount the cards.

Remember nVidia also makes AI accelerator technology and that is most certainly doing well, likely responsible for the inflated price of nVidia stock. And right now nVidia stock is definitely over inflated, buying right now comes at a 50%+ premium, especially in this bull market.

As for Intel Arc, a year ago I would of agreed there weren't a serious player. That changed this year when they got some really needed fixes in their drivers resulting in 11~50% performance increase. The Arc A750 is in the $200~250 range and works incredibly well at its price point. Their next generation release should start eating into the entry / midrange markets.

wow Saev you want a professional gig as a contributor?
you know I just got free perspective from you that I will be regurgitating tomorrow to our CIO (I'm the chief strategy guy) I argue all day long with the Chief investment guy.
To him and the CEO, A.I is just the next buzz word.
Totally blindsided by their bias of having experienced some hurt in crypto and metaverse. (both of which i told them were bad ideas. Crypto because of what I know of politics and history, the powers that be will never hand over control of currency or allow it to be liberated beyond their ability to tax, control. and metaverse because I am a gamer and knew that ***was a big old pile of dung.)
I'm also betting against Apple. forgot the name of their recently released product. the VR/AR big ol clunky Oculus-looking headset.

I am a strong advocate for how AR will be huge over the next few years. But I think Apple's approach to it, hardware-wise, is totally off. while their ecosystem (app store for VR/AR) is sound, it all is heavily dependant on access and adoption.
I can afford the 3500 USD headset. But I'm not about to pay that for what they're offering. and I'm also an android user and hate apple's customer experience.
"Pay for apple tv. its available to everyone. you dont need an apple phone or ipad, until you forget your password. Then we're going to make it hell for you to get back in. Unless of course, you choose to buy an ipad or iphon and just use them as your log in verification"
*** you apple and *** apple tv. I'll bootleg this ***just to take money away from you

AR is going to be really important but the hardware simply hasn't caught up yet, and likely won't for another decade or more. We need to develop wetware and human to machine translation first, then we can move the interaction from a clunky external device to electronics imbedded inside the consumer. We are a lot closer then people think, lots of development inside the biomedical field with prosthetics. Once we get better at building functional prosthetic eyes, then it's pretty easy to insert a digital overlay on top of the image. The guys working on functional prosthetic limbs are getting closer to having a limb that responds to signals detected from the remaining intact nerves. Once that happens we can interact with those signals and relay them the brain or other digital system.

AI is really just extremely good data modeling with clever mimicry. There is no actual intelligence behind it, the greatest achievement is that LLM mimic's how the human brains process language. They don't know why it works, only that it does. Makes for good Human Machine Interface (HMI), I expect this model will be further developed and become the standard HMI for digital assistants. Kyle Hill did a really good job of explaining it.

Ah cyborgs soon. Only if skynet doesn't come first.

By AR I meant augmented reality in the sense of visual overlays on devices. That's already here. Just poor applications, much like what Apple just released. There are some great applications that have been in existence for a while, like the smart navigation systems in some high-end vehicles like audi, bmw, and Mercedes. I found out, for example, that most of these vehicles' navigation systems are on-prem A.I feed back visual triggers like speed limits to update the displays. I always thought this was done via some kind of wifi/internet link to a centralized center where its fed back to the vehicle's system.
That kind of visual recognition and A.I must indicate we aren't too far from really good AR wearables.
That very much excites me because it will take virtual assistance to a new level. Combine it with things like real-time audio translation from one language to another, and we are stepping beyond star trek territory and fantasy, if we aren't already.
Furthermore, just imagine combining these wearables with the likes and capabilities of things like Microsoft's Copilot.
I think we're about to experience a great jump in productivity, efficiency and output. Greater than the disruption we experienced with the ICT revolution of the late 90's and early 2000s.

That is what I was referring to, except why bother with some external device when it's easier to just overlay on top of the signal from the optical nerves.

People researching functional prosthetic eyes are already doing this, though it's rather low resolution and grainy.

I'm just squeemish I guess. I'd never try to directly interface my biology with external tech.

On your point, not surprised honestly. I recall back in 2016? there were some tech that was able to translate brain scans of people in their sleep into visual images, to replicate what the individuals were dreaming of. they then did the same with people who were awake, asking them to visualise/imagine something. Have the machine interpret it based on their brain activity, into an image. had remarkably high accuracy.
however, it wasn't something that could be mass replicated because as they soon found out, different individuals had different patterns of brain activity.
BUT, as a result of this, correlations between language, thought and perception were discovered.
anyways I digress.
Tl;dr: I'm all for being a centurion, M.A.S.K, robotech type of cyborg.
But Bionic 6, Million Dollar Man, Robocop, no thanks.
I hear pacemakers can be hacked. Just imagine tomorrow some random ultra-hardline government decides its going to shut down the batteries of citizens who disagree with their policies. ft

Well proper biotech would be powered from the natural electrical current all humans produce, no batteries involved.

yeah just not for me. But all the power to them. Especially the prosthetics.
Cleveland clinic have a very advanced R&D center and great tech partnerships. I know they've been doing a bunch of prosthetic body augmentation for a while but nothing I saw was on the level of prosthetic eyes.

I've had enough experience with the Healthcare industry to know that I don't want something implanted in my body that interfaces with my nervous system

https://www.cleveland.com/healthfit/2018/01/former_cleveland_clinic_doctor.html

Pretty funny ***honestly. Imagine having a twin brother in this kind of situation.

No 50-series until 2025 and, unless the market changes substantially before then (like Intel actually becoming competitive in the enthusiast segment, AMD getting their ***together with drivers or something even more global), it's going to be priced just as terribly as the 40xx. Nvidia is not incentivized to compete on value because right now it simply has noone to compete with.

Personally, I got the 4090, but that's because I have too much disposable income and not enough common sense.

Intel is already making major strides in a lot of the areas that nVidia is trying to target. AI Upscaling for example: XeSS already shits on and blows away FSR which is really *** sad for AMD. I for one welcome the XeSS overlords though. It probably won't match bleeding edge DLSS versions, but it can be used by any card and is easier to implement in games. If they put out cards that can target the enthusiast market and not just entry level, they'll be cooking with gas.

No matter the technology, XeSS, DLSS or FSR, until it works without specifically being supported by a game, I will not base my buying decision on said technology.

I bought AMD (7900 XT) this time around, simply because it was the best option for the price I was targeting. If Intel had GPU closer to the top end, I would have probably considered it (considering the driver gains that Intel has been making). nVidia is seriously just ignoring anything but the absolute top end. You can get a 7900 XT for the price of a 4070 (non-TI) right now... its not even close.