News:

Willkommen im Notebookcheck.com Forum! Hier können sie über alle unsere Artikel und allgemein über Notebook relevante Dinge disuktieren. Viel Spass!

Main Menu

Post reply

The message has the following error or errors that must be corrected before continuing:
Warning: this topic has not been posted in for at least 120 days.
Unless you're sure you want to reply, please consider starting a new topic.
Other options
Verification:
Please leave this box empty:

Shortcuts: ALT+S post or ALT+P preview

Topic summary

Posted by NikoB
 - December 07, 2023, 17:40:41
up theme...)
Posted by NikoB
 - December 06, 2023, 19:45:54
Quote from: 02nz on December 06, 2023, 18:29:41typical consumer with a 4500U might well look at the 7520U and think they're going to get an upgrade, since it's "3 generations" newer and still a Ryzen 5, when in reality it's using the same architecture and has 2 fewer cores!
4500U - 6 сores. 7520 - 4 cores, but with HT - 8 threads.
Single core performance +3-4% for 7520, but multithread +15% for 4500U.

7520 support only LPDDR5 5500 (max 16GB onboard, no upgrade in slots). 4500U support only DDR4 3200 (LPDDR4-4266 onboard), but 64Gb MAX for 2 slots.

By the way, in order not to p*** off meticulous users, AMD has already removed the page with the 4500U from its website, so in some ways Intel is of course right that AMD are also scammers.

Crooks accuse crooks...
Posted by A
 - December 06, 2023, 19:31:13
Quote from: 02nz on December 06, 2023, 18:29:41when in reality it's using the same architecture and has 2 fewer cores
These cheap plastic low-performance laptops aren't purchased to calculate nuclear reactions. User will not notice 2 missing cores in Word and Excel. No one pairs those with good hardware and/or build. And most probably people buying these have no idea what "Zen2" is.
Posted by 02nz
 - December 06, 2023, 18:29:41
No question that most of AMD's mobile chips (esp 7040) are way better than Intel's at the moment, but it's also hard to deny that AMD pulled a fast one with the R5 7520U. You have to go back to the 3000 series to find another R5 with fewer than 6 cores. A typical consumer with a 4500U might well look at the 7520U and think they're going to get an upgrade, since it's "3 generations" newer and still a Ryzen 5, when in reality it's using the same architecture and has 2 fewer cores!
Posted by NikoB
 - December 06, 2023, 16:04:19
Intel has nothing to defend against even Zen3, not to mention Zen4, whose days are already running out.

They will have power efficiency improvements in Meteor Lake, but there will be virtually no performance gains - which was quite obvious from the start - or power efficiency improvements with the legacy "7nm" or speed increases with the same monstrous consumption. And the second is precisely what causes massive dissatisfaction among customers under heavy load. At low loads, thanks to better core optimization in collusion with Microsoft, they do much better than AMD, which is why they generally sell.

To simply match AMD in multi-threaded performance at 1W, they need exactly the same process technology or very close to AMD on TSMC. In the quantities of processors/SoC that Intel sells compared to AMD (5:1), it is obvious that even a theoretical complete, forced, temporary transfer of all production to TSMC is impossible for them - their volumes are simply physically unavailable at TSMC factories, where almost all conveyors ranks Apple/Qualcomm for the most advanced technological processes. AMD can play this game - and even then it is not interested in increasing supply volume in order to keep prices high with increased demand for its products, with its negligible market share compared to Intel, but Intel has no choice but to produce most of its products in its factories , into which a lot of money and effort have been poured.

Therefore, Intel today has only one true strategy to survive in the new world - becoming like TSMC, giving the opportunity to a lot of contractors to make chips at their factories - but this is an extremely unprofitable idea in reality, taking into account its status as a catching up and, moreover, increased competition in the contract market (unless, of course, we exclude oligopolistic collusion with TSMC/Samsung, which I definitely wouldn't discount in the modern world). At the same time, trying to maintain the profit margin from the production of our own processors and SoCs for laptops in order to stay afloat, while trying to maintain the level of technological leadership in terms of the real efficiency of x86 cores (if their efficiency is normalized not only by consumption, but also by the technical process - then I'm afraid AMD/Apple still turn out to be the losers for the most part).

This subtle process of "tuning" the company should ensure, taking into account, of course, subsidies from the authorities, its passage through the eye of the needle of current problems. But here it begins to very much overlap in interests with another strong company - Apple, and Qualcomm is already breathing down its neck.

In the market of such monsters, the winner is the most resilient (in terms of finances, human capital) and/or the one whose products are the most important, from the point of view of geopolitical interests. At the moment it is still Intel/AMD rather than Apple. But if the world starts to quickly move to native Arm code, in terms of code base, things will be very bad for them, in favor of Apple/Qualcomm and possibly Samsung/Mediatek. There is still a lot of jealousy from the support of the authorities and the desire to save the right company.

From the point of view of the average buyer, 2024 will definitely be boring. Even more so with an expert one. Based on the actual state of the shelves and the selection on it.

After such a sharp jump in performance for the Zen4 HX series, you shouldn't expect anything like this in the next 2 years, which means there is no particular desire for Zen4 HX series owners to upgrade. Unless for those who are avid gamers, and then only due to the fact that the mobile 4090 is about 1.5 times weaker than the desktop version, i.e. The rate of obsolescence of mobile DGPUs is much stronger than desktop ones, even though the 7945HX can easily service the future 6090, but in a laptop you cannot change the discrete chip yourself, which forces you to upgrade much more often than the processor becomes obsolete.

This would be especially obvious if the Zen4 HX memory controller were four-channel (or designed for at least DDR5 8500), but alas, it is significantly slower than that of the Raptor HX series. And it would be even better if, in HX chips, AMD decided to switch to HBM3 RAM (albeit with real bandwidth reduced to 300-400GB/s at the cost of a lower frequency or bit depth). Then it was a really balanced solution in terms of processor and RAM/memory/peripheral support for many years, but alas. And taking into account the 28 pci-e 5.0 lines (which, most of them, just hang in the air on the Zen4 HX, not being used in any way by laptop manufacturers, precisely because of the extremely slow RAM), it would be possible to display a 16x pci-e connector on gaming laptops 4.0, which uses only 8 pci-e 5.0 lines (but you can also install x16 pci-e 5.0 - all the same, 12 5.0 lines are available to other devices) and which would easily ensure an upgrade of such a powerful laptop in the future to 6090 or even 7090.

But we have what we have. Pointless 28 pci-e 5.0 lanes in 2023 laptops with Zen4 HX, most of which are just hanging in the air...

And Apple completely disappointed with the real bandwidth of its laptops compared to the declared bandwidth. They stated 300-400GB/s in the M3 Max (as in the M2 Max), but the real efficiency of the memory controller is only 120/300 - 30%-33%. This is literally 2 times worse than the efficiency of the memory controller of AMD chips, which themselves are inferior in efficiency to Intel memory controllers in terms of absolute efficiency in %.
Posted by JohnIL
 - December 06, 2023, 15:15:12
This is clearly the pot calling the kettle black. Either this marketing team was completely clueless to Intel's own misdirected naming schemes and repeated refreshes. Or they chose to ignore all that hoping most would forget.
I think Intel didn't move the bar much on their new architecture. Not that its bad or anything but certainly its nothing that will shake up things. Is saying that some of AMD's claimed new CPU's are just refreshes wrong to point out? No, but Intel could have taken the more level approach to marketing and simply provided good facts.
Instead they created a presentation that seemed more disgruntled and petty over something they too are guilty of.
It was correct to pull this terrible marketing ploy but will remain a black mark for Intel.
Posted by NikoB
 - December 06, 2023, 10:03:17
QuoteIntel was able to show the Core i7-1355U outperforming the Ryzen 7 7840U. Taking a broader view sees the latter push ahead of the former by an average of 10% though
The author probably drew conclusions based on an alternative reality. In those test results, key tests for heavy floating point loads, I see the 7840U having an advantage of more than 35%.
And that's without taking into account that the 7840U can maintain this performance at lower consumption indefinitely, while the 1355U has peak multi-threaded performance at which it quickly succumbs to lower values at exactly the same consumption. like 7840U, i.e. the real loss of the 1355U (as the tests in laptops prove) loses by more than 40% almost everywhere, except for single-core tests (pulse performance) and tests where the emphasis is on memory controller performance, where Intel is still in the lead as 5 years ago and as I have mentioned more than once was written.

Thus, except for tasks where the emphasis is clearly on memory bandwidth (for example, Photoshop), the 1355U is outperformed by the 7840U.

And the only reason why Intel is not yet bankrupt and is still sold everywhere is because they have their own factories and a market share of 5:1 compared to AMD, which is not worried about real market saturation - because it does not want to increase its market share over 25%, which has been proven over the last 8 years. Those. AMD is essentially a hidden anti-monopoly layer for Intel, their secret formal protection from anti-monopolyists and thanks to secret cross-licensing schemes (which Intel refuses to provide through anti-market methods to other companies like Apple/Qualcomm/Mediatek and other players they previously suppressed, like NVidia - remember , how did they block their right to produce chipsets for x86? And what did the corrupt US antitrust officials do? But nothing!) and keeping AMD afloat in its weak years, Intel created and is creating a fictitious appearance of real competition in x86, but in reality with a share market 5:1 in favor of Intel, it is obvious that this is its monopoly market, and AMD will always remain on the border of profitability and market share, exactly as it is convenient for Intel. Until civil society (if you can still find one in the US) forces politicians to make real competition.

Well, I hope that the days of both Intel and AMD are in the past, because... Apple clearly has many times more money; in fact, it controls most of TSMC alone. Although the technical progress itself in silicon-based IT is clearly quickly approaching a fundamental dead end.

So the survival of all three depends on some upstart in IT, who will suddenly offer something new, orders of magnitude better than them, although taking into account the costs of development, in the modern world this already looks like pure fantasy - rather, the one with the most money will win here and who will be the first to buy all the developments and patents of this new "super-upstart". If this "super-upstart" is not on the horizon, then the entire IT industry, in the next at least 30 years, will be a rotten swamp, gradually reaching a complete dead end with the increase in processor performance per 1W of consumption.
Posted by Superguy
 - December 05, 2023, 21:46:39
Quote from: Poster on December 05, 2023, 00:21:20I was shopping for my first PC build at the time of Intel's P4 when AMD had wiped them with their 939 socket platform. Intel,from Pentium 2 thought that all they needed was to increase the MHz year over year to constitute as an upgrade. AMD then and now showed them real improvement.

Let's be real here. That was both of them.  The big race was to 1 GHz and clock speed was everything because that's all people know. 500 MHz good, 600 MHz better. Both companies were trying to put out faster and faster, until AMD beat them to 1 GHz.  Intel put a hurried launch on the 1 GHz P3 but it was vapor ware.

The P3 introduced a lot of improvements, in particular SSE which sped up processing quite a bit.  The early P3s weren't much of an upgrade over the P2, but the Coppermines and later Tualatins were a significant upgrade.  The Tualatins in particular were destroying the Willy P4s, especially when overclocked.

The P3 core was much more efficient than the P4 and was actually scaling well, though it didn't ramp up the GHz as much or as quickly as the P4.

Intel was making a lot of mistakes during this time, especially ramming Rambus memory down everyone's throats. It was super expensive and no one wanted it. It also was a terrible match for the P3. Intel tried to kludge SDRAM for it with the i820 chipset and its memory translator hub (MTH), but that was a disaster (both in performance and failures). It eventually came around back to SDRAM for the P3. Rambus did ok with the P4 but it was still too expensive.  SDR killed performance on the P4 and DDR was good enough. Still, the damage was done and Intel wouldn't really recover until Conroe.

AMD did its first round of trickery with its naming here. Athlons were more efficient but didn't have the clocks the P4s had. So AMD did this PR (performance rating) crap to show the "equivalent" performance of their chip. Reality it was a price rate as it allowed them to jack the prices of their chips higher than they would have been able to otherwise. I don't fault them for trying to get as much as they could out of it, but let's not pretend that this crap is new.

I think the general naming scheme both went to made it more clear where each processor stood in the pecking order without focusing on the clock speed. I think it simplified things a lot.

QuoteI do agree that AMD naming scheme was created to cause confusion from the normal trend of the advancement of the first digit in the generation, which all consumers had been accustomed to. But they declared it more than once and never denied it.

This isn't limited to AMD or Intel. Everyone does it to get rid of stock of older chips. Maybe tweak them slightly and sell it as a refresh. Nvidia is particularly bad with this in the 7x0 and 8x0 chips and the MX series. They're even doing it now with the Super editions coming out early next year. They're milking the 4000 series for all they can, and the consumers are suffering for it.
 
QuoteNow Intel...
Hitting back when you feeling heat cause you have nothing impressive to show for your 14th gen and even though AMD purposely not increasing chips to cover demand for laptop market, they still giving Intel headaches.

Feeling heat, for both companies, is a good thing. Things get stale and more expensive when one can't apply the heat. I like to see both a healthy and competitive AMD and Intel. I just wish the GPU market were the same.

QuoteWe'll see what the real benchmarks for all 14th gen CPUs and AMD's 8050 Apus are in a few months. All marketing fluff in the garbage!!

Agreed. There's too much hype and BS floating around now to know what the real story is. I'm open minded about both and hope they both do well.
Posted by Superguy
 - December 05, 2023, 21:18:35
Quote from: indy on December 05, 2023, 05:25:45All I see is fear. Leaders of a product segment don't look backwards or to their competitors.

Are you for real?  Of course they have to look back at their competitors to see what they're doing. Both AMD and Intel have gotten their clocks cleaned by the other when one got fat and happy with their current products only to be caught with their pants down when the other put out a killer product.

It happened when AMD punked Intel with the Athlon and the Hammer processors, and when Intel punked AMD with Core2 and Nehalem. AMD was relegated to the low end until the Ryzen's came up and started killing it.

One ignores their competition at their own peril.
Posted by JohnIL
 - December 05, 2023, 14:15:09
Intel has had its share of regurgitated stuff as well. Not sure they actually moved the bar much with their hybrid mess. Only made it more complex in order to claim some sort of edge. What Intel fails to realize with AMD is that sticking with what works isn't a bad thing. If you cannot really improve anything why bother. Intel certainly went years playing this if it works don't fix it with Skylake. Marketing today seems to focus more on tearing down your competition. That in itself is admission of your lack of faith in your own product. 
Posted by A
 - December 05, 2023, 11:25:58
They have such a long story of sh*tting each other, yet there are people who are seemingly surprised every time.
Posted by Hotz
 - December 05, 2023, 09:46:39
Quote from: LostInSpace on December 04, 2023, 21:26:40if AMD wasn't in the game we would still have single core Pentiums running at 500 MHz, with sockets for a FPU and GPU that cost $500 each. ;-)

Maybe.

But then also games would be much more optimized, because otherwise they wouldn't work. Nowadays developers don't care about optimization anymore and you'll need 20x higher hardware specs to get the same graphics as Crysis in 2008.


Just saying... faster development doesn't necessarily mean it's better.
Posted by TimApple
 - December 05, 2023, 08:45:59
Good Morning,

Just switch to a mac for the breathtaking experience of world's most advance 3nm M-series processors.

Yours Truly
Tim Apple 😁
Posted by indy
 - December 05, 2023, 05:25:45
All I see is fear. Leaders of a product segment don't look backwards or to their competitors.
Posted by George
 - December 05, 2023, 00:58:49
More like 'Core 1/2 Truths' !!! :)

Frankly I think both companies tend to flood the market with products with mixed heritage all in a vein effort to make more marketing hype (noise) then the other.

Then again, is it to much to ask for greater than ~%15 improvement[/i] in cherry picked benchmarks generation over generation?

(maybe +%30 overall and NOT cherry picked?)


one has to wonder...