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

[The Mass Shooter Ron Soye]

By the way, HUB's review of the 9600X has been posted and this comment really sums up the current CPU mindset:

Intel and its' consequences have been a disaster for CPU benchmarking.

I'd say part of what hurt AMD here is going too low, so all the techtubers are sobbing about no performance gain.

If it was, I dunno, 80 watts instead of 65, then that would still use less power, and also have a performance boost.

Or... the techtubers could do an efficiency/watt thing. but that requires integrity. And intellect.

All they have to do is compare the Ryzen 7 9700X to the Ryzen 7 7700, both having the same 65W TDP and roughly the same power consumption. I can easily cherry pick to show that the new 8-core often has low gains or loses against the old one (where you would expect +16% on average) and it seems power constrained when under load. Some reviews like Phoronix's are much more positive for the 9700X. Did Larabel choose a better mix of workloads, is Windows the problem, or what?

There were stories like this in June:
AMD Might Re-Spec Ryzen 7 9700X “Zen 5” CPU With Higher 120W TDP To Outpace Its 7800X3D Chip

If there is any truth to that, AMD might have noticed what we can see now, that the 9700X needs more power to shine.

Compare to the Zen 4 launch. AMD launched a 7700X (105W) in September 2022. Not naming it the 7800X tempered expectations (especially compared to the 5800X3D) and ensured that enthusiasts were looking forward to the 7800X3D. The 7700 (65W) launched in January 2023. The 7800X3D launched in April 2023 with a 120W TDP (!) that it doesn't really need in gaming, since it often uses closer to 50 Watts.

This time around, we have the 9700X at 65W. It's highly likely that a 9800X3D will be at least 105W, and with full support for overclocking for the first time it may handle higher voltages and actually reach that. Previous X3D overclocking restrictions were due to voltages that the cache die could reach, not heat.

AMD could have called it a 9700 instead of a 9700X. Unforced error. They may be sandbagging this one to make the 9800X3D look better and become the top seller at a higher price. I would prefer for X3D and non-X3D to launch at the same time.

AMD should scrap their TDP method anyway since it's so loosely rated that it's pointless, especially since most models don't come with coolers anymore.

AMD and Intel have both had arbitrary methods of defining TDP but the AMD chips show consistent behavior between generations:



The Package Power Tracking (PPT) limit is 88W for 65W chips. I think it's always 35% higher than the TDP (170W TDP = 230W PPT).

 

[Fapcop]

I'd say part of what hurt AMD here is going too low, so all the techtubers are sobbing about no performance gain.

If it was, I dunno, 80 watts instead of 65, then that would still use less power, and also have a performance boost.

Or... the techtubers could do an efficiency/watt thing. but that requires integrity. And intellect.

There IS a performance gain though. It's not massive, a solid 10-15% on top of using less power, which is a real achievement.

Techtubers are just chasing outrage clicks though. "OHNOES AMD IS ALSO FUCKED!!!" gets more clicks than: "Solid update! Less power, better performance!"

EDIT: Meanwhile it seems that Intel continues their retarded game of JUST THROW MORE WATTS AT IT!!!

https://wccftech.com/intel-arrow-lake-panther-lake-power-ratings-top-desktop-sku-295w-pl2-canceled-40-core-die-352w-pl2-667w-pl4-tdp/

 

[geckogoy]

There IS a performance gain though. It's not massive, a solid 10-15% on top of using less power, which is a real achievement.

Well on multicore its matched or slightly outmatched by previous gen due to the aforementioned power limit

 

[Fapcop]

Well on multicore its matched or slightly outmatched by previous gen due to the aforementioned power limit

Maybe in some cases, but clearly not in all:

 

[geckogoy]

Maybe in some cases, but clearly not in all:
View attachment 6289285

I'm just saying, its more like -1% to 15%

They could of thought out the power limits a little better and probably gotten a far better showing

 

[The Mass Shooter Ron Soye]

My pet techtuber said some of the same things I did and more:

Spoiler: TL;DW

• Previously told gaymers "Do not wait for Zen 5, just get a 7800X3D".
• AnandTech's SPEC2017 test is in line with IPC uplift that AMD marketed for Zen 5, Tom's Hardware review shows gaming uplift, Phoronix review shows uplifts.
• AMD botched the launch. "This was one of the biggest self-owns in history." It should have been delayed a few more weeks.
• "Whenever you see a launch get way better results in Linux, a good example is early Zen[1] Threadripper reviews when they hadn't quite figured out the scheduling, then you know that there's something wrong with how the software is utilizing a new architecture in Windows."
• PCWorld delayed their own 9700X/9600X reviews due to inconsistencies/anomalies in their own testing.
• The chips are held back by power, voltage, and bandwidth constraints. Linus Tech Tips found that the 9600X was hitting higher peak boost clocks than the 9700X, showing that 65W isn't enough for 8 Zen 5 cores to "stretch their legs". Also applies to single-threaded performance since peak clocks on the 9700X were lower.
• Giving it a higher TDP/PPT lets it match the 14-core 14600K / 12-core Ryzen 9 7900 in multi-threading.
• AMD should have given reviewers more guidance on FCLK/memory tweaking, and shipped faster memory kits to reviewers, which Intel does. It can clearly go above DDR5-6000 and benefit.
• AMD should have given the 9700X a 105W TDP. It could have matched the 7800X3D in gaming and 7900 in multi-threading.
• They should have launched these on the same day as the 9950X and 9900X, so people would have focused on the 9950X's better single-threading performance.
• All the chips should have launched at the end of August. AMD is rushing Zen 5, Strix Point wasn't quite ready at launch, the 9700X and 9600X have inconsistencies in performance.
• Delaying until the end of August would have given them time to put together more impressive review kits with faster memory, better overclocking guides, more answers to when you should tweak the memory or FCLK.
• AMD screwed up by not announcing the release date and pricing of Zen 5 X3D. All people know is that they're disappointed by the 9700X, not that there is a 9800X3D coming in a month or two that is 20-30% faster.

Imagine how different reviews would have looked if AMD did those things:

1. The 9700X launches with a 105W TDP, and therefore matches/narrowly beats the 14900K(S) in gaming, and matches the i5-14600K in multi-threading.
2. Reviewers would have praised Zen 5's efficiency in 65W ECO mode, but also praised the performance at stock settings.
3. Reviewers would have noted that faster memory kits (DDR5-6400) would boost performance even more and extend AMD's lead.
4. Because (in this scenario) the 9950X launched on the same day, nobody would have cared that the 9700X (at 105W) is only slightly faster than the 14900K in gaming 14600K in multi-threading, because the 9950X would be another 10% faster when using PBO, making it the best flagship.
5. Everyone would have imagined how cool the (already announced in this scenario) 9950X3D would be. Instead, "nobody is excited for the flagship anymore", and nobody's mentioning X3D because AMD is being "perplexingly quiet" about it.

How does AMD unscrew the launch?

• Immediately announce the 9800X3D before the 9950X even launches, including price and vaguely how it will perform. Maybe I will be proven wrong, but I expect 9000X3D to be 15-30% faster than the 7800X3D. People are going to be surprised by the uplift.
• If the 9800X3D is ready to launch within a month or two, the 9950X and 9900X should be delayed to launch on the same day. This would give AMD time to ensure that software and BIOSes are extra polished for when the actual gaming and multi-threading flagships, the 9800X3D and 9950X, actually come out. It would ensure that on review day, everything is portrayed in the best light in every direction. You'll have 9950X with PBO charts being shared at the same time as 9800X3D, showing all of the best aspects of Zen 5.
• If X3D isn't ready to launch for several months, AMD should work to launch a 9800X that boosts to 5.7 GHz with a 170W (!) TDP. Delay the 9950X, because you need to show a flagship gaming chip next to the flagship multi-threading chip at the same time. It would give time to ship with the latest X870 motherboards and the best RAM possible.
• I do not recommend the 9700X and 9600X. The 9950X is probably the only product I'll like out of the lineup.
• Intel doesn't know how to not be #1, AMD doesn't know how to be #1.

You can blame Intel and the consoomers for not appreciating lower power chips, but there are other issues beyond that. The launch was already delayed due to chips not meeting quality expectations, and it's only conjecture that bad silkscreening (e.g. Ryzen *9* 9700X) was the only problem. Gamers Nexus still ended up with a bad 9600X, and reviewers are finding inconsistencies in performance (mostly in Windows, but I can find examples of the 7700 non-X beating the 9700X in Phoronix's review). The 9700X is half-baked and needed more time in the oven.

 

[Shao Khan]

I've complained before that the 12 GB of RAM in the Radeon 6700 XT never seems to be useful. A recent example is No Man's Sky. With everything except textures on Ultra, it stays under 8 GB and runs at 90 fps. Bump up the textures to ultra, and now it's consuming 11.5 GB and running at 30-40 fps inside the hangar of my capital ship.

With resident evil villa and 4 everything cranked up to 11 totally max outs the ram. What the fuck does no man sky to need that much power?

 

[The Ugly One]

With resident evil villa and 4 everything cranked up to 11 totally max outs the ram. What the fuck does no man sky to need that much power?

High-resolution textures. You can max out the ram in Doom if the textures are high-res enough.

 

[WULULULULU]

For undervolting + overclocking a 6750XT, are these good? Using Linux as an OS so no AMD command center.

 

[The Ugly One]

This doesn't occur on the 6750XT... right?

A 6750XT only has 12.5% more bandwidth than a 6700XT, and less than a 3060 Ti, so I wouldn't expect the situation to be much better.

 

[Fcret]

Splitting the launch of Zen 5 seems like such a confusing decision to me. No one is absolutely clamoring to get the 9600X and 9700X. At best, kids with constrained finances will pick these up to game on a year from now.

AMD has more or less intentionally poisoned the first week of the Zen 5 release news cycle and I just don't understand why.

 

[The Ugly One]

Splitting the launch of Zen 5 seems like such a confusing decision to me. No one is absolutely clamoring to get the 9600X and 9700X. At best, kids with constrained finances will pick these up to game on a year from now.

AMD has more or less intentionally poisoned the first week of the Zen 5 release news cycle and I just don't understand why.

There's nothing intentional about it. They have a quality issue they discovered way too close to launch. When that happens at a large company, chaos ensues. Conference centers are already rented out, advertising contracts are already signed, etc, everything becomes a shitshow.

 

[seri0us]

For undervolting + overclocking a 6750XT, are these good? Using Linux as an OS so no AMD command center.
View attachment 6289577View attachment 6289579

No clue, but what program is that?

 

[AmpleApricots]

I'm happy with the 4650G Pro in my desktop.

Problem is, it's finally time to upgrade my wife's 1400-based system. I've been thinking - I swap my 4650G Pro into that computer (Mainboard should support it w. update) and then I can also get rid of her ancient r9 390. (That system really is just a frankenstein's monster of my old parts; she doesn't play vidya besides a bit of stardew valley maybe the iGPU of the 4650G can handle easily) I wanted to redo that computer several times but she's not a tech person and being weird about me changing her computer. Also windows it's gotta be, her work software is obscure and deeply buried into windows APIs and simply needs windows.

..OR I could also update to AM5 and give her my entire computer as is wholesale and buy something new. It doesn't really feel like it's worth it. That performance is kinda wasted on me. I don't really care about FPS much and all I play if I ever do are CPU heavy games like stellaris, Crusader Kings, factorio (barely) or rimworld. If I (rarely) play a 3D heavy game, I just limit framerates to 30 FPS. I do not care. That said, people seem to be in love with the 5800x3d as drop-in replacement for AM4 computers, especially for compute heavy games like the ones I listed because of the big cache. In my corner of the world, that chip costs about 300 Euros. That seems kinda expensive for an old chip though, and I also would hate to lose an iGPU for the power efficency gains but might be able to stomach it, especially when I update my graphics card somewhen, seeing as I do most of my non-heavy crunching computing from a simple 5W TDP CPU thinkpad anyways. Thing is, I might also want to move over some RAM to my wife's computer and it feels kinda stupid to buy DDR4 RAM which in my corner cost almost as much as DDR5 RAM. It might be better in that case to just swallow the bitter pill and upgrade to AM5.

..OR I buy her a macbook. She'd like it I think, downside is that that macbook would not last nearly as long as the old Zen and don't tell her I said this but she breaks absolutely everything. Yes, she's one of those people. That computer lived as long as it did because she can't carry it around. Also her software probably won't run on the macbook either. Maybe with virtualization, but I have no idea about this in regards to Mac software. I don't feel this is a very viable option.

so do I eiter A) move the CPU and buy a 5800x3d for myself (or some other AM4 CPU) B) Buy an AM5 system?

 

[WULULULULU]

No clue, but what program is that?

LACT.

 

[The Mass Shooter Ron Soye]

so do I eiter A) move the CPU and buy a 5800x3d for myself (or some other AM4 CPU) B) Buy an AM5 system?

I say try A) but get a 5700X3D instead if that's the better deal (it is in the US).

You'll have to figure out your RAM situation. X3D can compensate for slower, cheaper RAM (in gaming).

Buy cheap laptops that are practically disposable. An Intel N100 isn't that much slower than a Ryzen 5 1400.

 

[Fcret]

There's like a weird pseudo-war happening between Linux/workstation users and gamers over the value of Zen 5 happening right now on Twitter and Youtube. I don't know why everyone needs to sound off right now when we don't even know what flagship is going to look like in the field.

 

[Just Some Other Guy]

There's like a weird pseudo-war happening between Linux/workstation users and gamers over the value of Zen 5 happening right now on Twitter and Youtube. I don't know why everyone needs to sound off right now when we don't even know what flagship is going to look like in the field.

Idk. Lots of dooming going on about how because AMD released a pair of more efficient chips that enthusiasts were never going to buy anyways, it must mean the entire product stack is going to suck.

 

[AmpleApricots]

but get a 5700X3D instead if that's the better deal (it is in the US).

Now you said something: That one costs about 100 eurobucks less here. For 200, it's more of a no-brainer.

 

[The Ugly One]

..OR I buy her a macbook. She'd like it I think, downside is that that macbook would not last nearly as long as the old Zen

Macbooks are the longest-lasting laptops, most durable I have ever owned, and I have had a lot of laptops. So has my wife. 6 years in a backpack, and her MB Pro is still in great shape. But if the stuff she needs doesn't have an OSX version, forget it.