Installer Steam
connexion
|
langue
简体中文 (chinois simplifié)
繁體中文 (chinois traditionnel)
日本語 (japonais)
한국어 (coréen)
ไทย (thaï)
Български (bulgare)
Čeština (tchèque)
Dansk (danois)
Deutsch (allemand)
English (anglais)
Español - España (espagnol castillan)
Español - Latinoamérica (espagnol d'Amérique latine)
Ελληνικά (grec)
Italiano (italien)
Bahasa Indonesia (indonésien)
Magyar (hongrois)
Nederlands (néerlandais)
Norsk (norvégien)
Polski (polonais)
Português (portugais du Portugal)
Português - Brasil (portugais du Brésil)
Română (roumain)
Русский (russe)
Suomi (finnois)
Svenska (suédois)
Türkçe (turc)
Tiếng Việt (vietnamien)
Українська (ukrainien)
Signaler un problème de traduction
Procédures à suivre et règles des discussions
Signaler ce message
We can blame bad optimization, but it doesn’t change anything; it's been the reality of PC gaming since the very beginning.
I would love to move up to 4k, but the problem is that even the best hardware still struggles with 1440p 144hz so its not really worth getting 4k just yet.
Maybe the 4090 can run some games at 4k 144hz, but i have a 4080 and it certainly struggles at 1440p.
Games upscaled from 1440p to 4K look very close to real 4K while having close to 1440p specs requirements.
AMD has created a monster.
I guess they will drop the next "monster" again when they introduce Zen 6 end 2025?
People act as if they've never seen processing improvement in their lives......
1. There are gains as high as 25~30% over the previous generation and the i9-14900K, even larger gains if you compare to Arrow Lake, the largest example being in BG3. Large improvements in 1% and 0.1% lows which means that gameplay in relevant titles is even smoother than it was before. And with games getting more and more demanding, the gap will only get more noticeable.
2. Unlike previous 3D SKUs, this is overclockable because they moved the cache underneath the CPU cores, so people that want to take the risks can push the gap even wider, or undervolt for better efficiency. The adjustment also means that it's easier to cool than the 5800X3D and 7800X3D, so you really don't even need really expensive cooling to run it at stock or undervolted.
3D cache is a monster in of itself that has been terrorising Intel since its first iteration with Zen3, and it's only going to grow.
I hope Intel will cach up as currently there is no reason to buy their CPUs.
I always was an intel fanboy.
ever since the 90s the k5 and k6 and thunderbird days it was :
amd best bang for buck (10-15% cheaper for same performance) but less stable (much more blue screen of death)
intel a bit more expensive but more stable..
nether had a true lead in absolute performance crown.. it was do you value stability?
as one who hated my system crash-reboot in the middle of a 4 hour long game intel was my thing..
but today intel.. well intel needs to pull a ryzen to get back on track..
amd already starts pulling the same price increasing trucks intel did when all amd had were those crappy fx things..
I'm interested to see where things go with RISC-V, moving forward.
I did the same thing with a 3700X to a 5800X3D, and it was sometimes a massive uplift.
This, and even the original Athlon (Thunderbird) scored some wins against the Pentium III, especially later since the Pentium III stopped scaling sooner so the Athlon beat it to 1 GHz and was stable beyond too. Only the rare Taulatin Pentium III may have exceeded it, but since that needed a new board and was limited to boards with 512 MB maximum support, it wasn't a common CPU.
The Athlon XP also outclassed the Pentium 4 (Willamette and early Northwood), with the latter only reaching parity with AMD once Northwood came of age and gave it 512k L2 and 800 MHz FSB, which was short lived equality because intel would soon slip backwards with Prescott, and the Athlon 64 was coming too.
Even the K5 and K6 had their wins. A lot of stuff was often compiled for Intel back in the 1990s/2000s as well which could give Intel a manufactured advantage.
A lot of people think Intel was always on top until Ryzen but this is not so. It's more that they lost the top spot with Core 2, didn't regain it with Phenom, and then slid into a lengthy time period of setbacks with Bulldozer (the L1 cache issue hurt it bad, but AMD had neither the time or money to redesign it so it was literally make or break, and it was almost break for them).
The original Athlon wasn't Thunderbird; those were Pluto and Orion (same generation, two different tiers of base chip). Thunderbird was the 2nd generation of the Athlon.
K6-2 and K6-III both bested the equivalent pentium at the time (P2 333MHz vs K6-2 333MHz; and P3 450MHz vs K6-III 500MHz) especially when something wasn't compiled with the Intel compiler that specifically optimized for Intel. Intel's later P3's regained the crown until around their 933MHz offering, but there wasn't a massive lead. AMD took the first to 1GHz with the Athlon and regained the lead slightly. After that short regaining of the crown with the Athlon Intel again retook the lead with the P4 pushing the power and clocks with the P4EE models through Prescott when they were very close between those P4s and the Athlon 64 equivalent; e.g. around the Athlon 64 3500+ vs P4 3.2GHz. That whole era from Tbird Athlons through the Athlon 64 was one of the most competitive in the history of CPUs.
After that AMD was clearly the better performance due to the short comings of netburst with its 20 stage pipeline. It was incredibly high performance when software was specific and predictable; but it was terrible when anything tossed a curve ball and it had to flush the whole pipeline. As games advanced during that time pushing clocks higher just couldn't overcome the random stalls when an instruction needed to change and the whole pipeline flushed. Up until AMD got blind sided by the original Core Duo at the same time they were hitting their own wall after AMD Phenom with Steamroller Steaming pile of crap, Bulldozer ♥♥♥♥♥♥♥♥, and Excavator Excrement; They were definitely easily holding the performance crown. Intel taking what they learned from their issues with netburst and rebuilding from the OG Pentium Pro architecture to make Core pushed them way beyond where AMD was at and that ushered in Intel's most recent dominance for the better part of a decade or more.
Zen (1) honestly wasn't even all that good and it had plenty of architectural issues, core-to-core latency was terrible for example; but it was at least good enough to get them back in the race. Zen 2 helped mitigate things a bit but it wasn't until Zen 3 (Ryzen 5000) where they got to what probably should have really launched as Zen 1 if they weren't in the dire financial position they were in at that original Zen launch. Things have still been competitive but specifically for gaming, after the Ryzen 5000 series things are slightly skewed toward AMD on the CPU front; and with the 9000X3D changes with Zen 5 they are widely in the lead for games that are CPU heavy; sims (as in simulators, not "The Sims"), etc.
Hopefully Intel is over the apex of their struggles in their fabrication engineering and as they move into 18A process with ribbonFET / GAA nodes they can get back to being more competitive with AMDs currently offerings. I'm still somewhat doubtful with the recent rumblings that the original overhaul and new ground-up design that Jim Keller was leading before he left seems to have been completely scrapped. Probably one of the most boneheaded moves by the former Intel CEO was pushing him out the door because they didn't want to listen to him regarding their fab engineering failing and giving him autonomy to look outside intel for fab options.
But thanks for the history lesson, since that was a bit before my time and most of what I learned of it was retroactively and everyone always referred to the Thunderbird variant, so I figured that was the relevant one for what I was saying anyway.
I think Zen was a fair launch model. Yes, it didn't instantly put AMD in direct performance competition with Intel at the time, at least not in gaming or individual core performance, but it was a huge increase over what AMD had before, it was maybe around overclocked Haswell performance when Intel was on Sky Lake/Kaby Lake which isn't that far back, and they made up for it with a lot of cores. Yeah, yeah, "more cores" memes aside, people and software were really starting to want more than four cores by this point, and most importantly, the price was right. It was a great foundation, slightly tarnished by early issues on the BIOS side, and RAM compatibility/high speed support wouldn't be there for another couple generations. It probably took a while before motherboard vendors "trusted" AMD since they had just a low market share that none of them wanted to put much effort into an AMD board maybe? But they worked through those issues.
Zen+ was muted but Zen 2, despite still being a bit behind, really showed they were for real on threatening to catch back up to Intel. This laid the foundation for their chiplet strategy, although the split CCX per CCD hurt it a bit in gaming. But this "hurt in gaming" was still closing the gap on Intel. I was upgrading from my 2500K and almost defaulted to Intel because while I knew AMD had something promising going, I knew I also wanted strong core performance. However, I ended up choosing the 3700X over the 9th/10th generation offerings entirely on price ratio. I took Sandy Bridge over Bulldozer, but the difference was there was a fair performance gap there. From most of the results I found, the 3700X was within 10% of the core performance of the 9900K/10700K, but was 2/3 the price (or less if I account for the included cooler saving me a few months of up front costs on an aftermarket one). Being able to drop a 5800X3D into the same platform was just the sweetener.
I hope Intel gets competitive again because I'd rather see a repeat of the early 2000s than a repeat of the early and mid 2010s.
Concurred. I'd much rather have the era from P3 933 - Athlon XP than the bulldozer era; even though Intel in the first half of the Core era was still pushing themselves even though AMD wasn't. But for one reason or another Intel's fab engineering / materials talent lost their way. I don't personally buy into the "intel resting on their laurels" notion that is the common defacto reason given for the 14++++++ process node... I'm more inclined to attribute it to a loss of their real talent and a dilution of talent with their bloated teams; which is why I'm still not super optimistic for a turnaround in the near future.
And I would have to imagine AMD not competing, while it may not have been "the" reason, was still "a" reason as to why Intel may have prioritized shareholders over the future at the time. Then once Ryzen lit a fire under them, they had to panic and, at first, add cores, and then they had to start to push those cores even further when Zen 2+ started closing the gap, and that resulted in heat and power draw became bigger issues. Then the 11th generation saw a regression in some regards, and the 12th generation was actually a genuine improvement (well, in theory anyway... the socket bending, the scheduling issues causing games with anti-cheat/DRM from properly working, and the e-cores not always playing nice were real teething issues... but the performance potential was there and it was Intel's biggest uplift since perhaps Sandy Bridge). I'd even call the 13th generation (but not the 14th) a refresh that warranted praise, as it reminded me of the 45nm Core 2 refresh... at least until the problems with Raptor Lake surfaced. As it stands, the 10th generation/12th generation seem like Intel's best in recent times, and even those weren't perfect (10th was a bad proposition in my opinion until Zen 3 launched and forced Intel to price cut it [which then made Zen 3 a bad proposition], and 12th had some initial issues mentioned above that only got forgotten about once Raptor Lake's issues surfaced).
As for Arrow Lake... too early top speculate where this goes. It seems like it has potential in productivity but its current performance for price is just awful even for that (A Ryzen 9 can nearly match it in productivity, while being cheaper [and no need for fast CUDIMM RAM], and better in gaming), and gaming performance is down due to latency going up. Seeing numbers like "40%" and "60%" slower (referring to 285K vs 9800X3D in select games) is almost unheard of and reminds me of Sandy Bridge to Bulldozer. They claim to have improvements coming for gaming performance but we'll see.
But Intel's current issues don't even seem to go back to Raptor Lake's issues. They seem to go back MUCH further, when they allowed TSMC to get so far ahead. Intel is too important to be allowed to fail, so that won't happen, but things like spinning off their fabrication, or being bailed out the government (and perhaps force merged with another American company) aren't entirely out of the question. A few years ago, such possibilities would be met with "you're overreacting".
Times will be interesting. I just hope we continue to see two players feeding off of one another to fuel progress. All of this doesn't even touch upon ARM threatening x86 entirely. AMD and Intel might need to start working together more, as unlikely as that sounds.