• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Educate me and maybe sell me to Intel?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

don256us

Uber Folding Senior
Joined
Jul 17, 2003
OK guys. I'm in the market for a new gaming rig. Nothing fancy but building around a 3080Ti. I'm probably an AMD guy but I don't understand Intel CPUs and I'm looking for someone who knows better. Simple stuff here. I'm not looking to break the knowledge bank here.

P cores and E cores. What gives? Performance and economy? If a CPU has 6 P cores and 4 E cores, it gets listed as 10 cores. If they are different, how do I compare the performance of a "10 core" CPU to AMD or another Intel?

In terms of actual recomendations (if you have one) I mostly do Folding @ Home. Gaming is limited to Total War: Shogun 2, Civ VI and other lower tier games but FPS standards. I Fold, game a little game, stream YouTube mostly, email and web browsing.
 
Lots of reviews out covering this stuff when Intel released Alder Lake last year (our front page review to name one :)) , but you're essentially correct. One is a performance core that runs a lot faster uses more power. The e-cores use less power and clocked less, correct (there are more differences, architecture is).

As far as direct comparison core for core, you can't because AMD doesn't currently subscribe to the BIG/little that Intel is doing so you look at it as a whole. Look at the horsepower, 0-60, 1/4M times numbers (benchmarks results), not how many cylinders it has is my best advice. ;)

I'd imagine you're not folding with the CPU (especially these power hoovers), so I'd gather your only concern is with PCIe lanes and multiple GPUs? If you're not dropping in multiple GPUs, then I wouldn't consider F@H in your purchase as you have the slot for a GPU to crank out points and no other concerns (right?).

With Intel (assuming you're looking at latest/last gen), if you're budget limited, you can at least save some money and go DDR4 with the latest processors and chipsets. With the latest AMD, it's all DDR5 and boards are generally more expensive than their Intel counterparts.

What's your budget? What parts do you need? We can make better recommendations with additional information. :)

Also, you stream?! Awesome! What's your YT channel? Whaddya' stream/your content? Your NV GPU takes care of that anyway, so you're set on that front.
 
Last edited:
Come on man. ;) Give a guy a break.

I still think that I'm going AMD Ryzen 7 3700x due to price vs performance vs energy use vs PCIe lanes. I just need to make sure that I'm sure.
Post magically merged:

Lots of reviews out covering this stuff when Intel released Alder Lake last year (our front page review to name one :)) ,
OK. I went to look at the Adler Lake reviews I promise but got too engrossed with the Ryzen 3900/3700 article instead.

I think you may be right that in order to have enough PCIe lanes for a x16 GPU and two x4 NVMe, I need AMD?

I'm looking for Intel fanboys to correct me and show me the light.
 
Last edited:
I think you may be right that in order to have enough PCIe lanes for a x16 GPU and two x4 NVMe, I need AMD?
I didn't say that, lol!

Intel Z590/690/790 will have plenty of lanes for one GPU and two PCIe 4.0 x4 SSDs. If you go Z790, there's PCIe 5.0 x4 M.2 socket in most (all?) boards. In all cases this drops the primary graphics slot from PCIe 5.0 x16 to x8, but remember PCIe 5.0 x8 = PCIe 4.0 x16 bandwidth. This gen, nor the next-gen flagship, and likely not the next-gen will choke on 4.0 x16 considerably... take a look at THIS PCIe scaling article and how little a 4090 chokes at PCIe 3.0 x16, for example... ;).

You can also use the 4.0 x4 M.2 sockets attached to the chipset and leave the full x16 bandwidth if that's somehow an issue. Z790 has it's own 4.0 x4 off the CPU too. Look up "z790 chipset diagram" and look at how it splits off.
:)

What are you trying to spend on a this new mobo/cpu/RAM combo??
 
Last edited:
in order to have enough PCIe lanes for a x16 GPU and two x4 NVMe, I need AMD?
Kinda. If you have to have those NVMe drives on CPU lanes at the same time as x16 connected GPU, then only Ryzen 7000 can provide that with consumer tier hardware. Ryzen 5000 and older, and Intel, both only have 20 usable CPU lanes. Don't get tricked by AM4 having 24 PCIe lanes since 4 of those are reserved to go to the chipset. You get the same 20 usable lanes as Intel.

If you're ok with the 2nd one going on chipset lanes, you can go either way.
 
If you're ok with the 2nd one going on chipset lanes,
Serious question.....who isn't OK with this? Will you share the reasons a user would want/need two directly attached to the CPU? Are there performance differences you can see outside of a benchmark?

Correct me if I am wrong, but this was only a thing a month ago with AM4. (I don't count AIC cards in PCIe slots - IIRC, only a couple of AM4 boards support the bifurcation needed to do this), right? You're spot on... but unless someone has that curious requirement, a single GPU and two (or more) PCIe 4.0 M.2 sockets can be run on Z590+ without sacrificing PCIe bandwidth to the primary graphics socket.
 
Serious question.....who isn't OK with this? Will you share the reasons a user would want/need two directly attached to the CPU? Are there performance differences you can see outside of a benchmark?
I don't want to assume what someone may or may not want to do. It is a consideration that may or may not apply.

About the only scenario that might be interesting is if you heavily use devices attached to chipset. Maybe you're doing 4k60 capture over USB, and there's fast network connection going. If you then hit the drive hard, I don't know how bandwidth shortages are handled by the system, but it may glitch other uses.
 
Hmmm.... good food for thought. I probably don't care if the lanes are chipset or CPU based on your thoughts. I appreciate the heads up. I'm still thinking AMD for general cost past the CPU. MB's seem to be cheaper for AMD(?). Also, if I don't concern myself with single CPU specs, AMD run with less power which will help keep long term costs down if I go with a 65w CPU from AMD.

I've not seen Intel competitive in that area but maybe I'm putting too much into wattage?
 
MB's seem to be cheaper for AMD(?).
The last-gen boards are, generally, cheaper yes. Their newest stuff, X670E costs more than Z790 and Z690 though. I wasn't (still?) not sure what you're aiming for so I mentioned that earlier with the new products in mind.

With Intel CPUs, if you're looking previous-gen and mid-range, they're competitive (right @DaveB ?).

I've not seen Intel competitive in that area but maybe I'm putting too much into wattage?

AMD does generally run with less power at full tilt, indeed. But if you're gaming and putzing around...

...... the difference isn't as significant.


Also, I see you've mentioned 3000 series from AMD... but I'd go 5000 series at this point. IIRC, the 5000 series (outside of 7000 series, lol) was a notable jump over the 3000 series. Consider double-checking reviews and see if the difference is worth the cost. But a mid-range 12600K or something sounds right up your ally too....
 
I've not seen Intel competitive in that area but maybe I'm putting too much into wattage?
There's many factors but even for latest gens of both, it looks like Intel are still not as power efficient as AMD. However the gap isn't necessarily as big as some clickbait sites make out.

image.png.03350422d6e6880e0e425841ede654df.png

For example, here's a power scaling chart I made previously putting the 13900k against the 7950X. Horizontal is power, vertical is MT Cinebench score. Data was taken form Hardware Unboxed and Computerbase. HUB I believe have since retracted their results as they discovered the mobo they used to test with supplied excessive voltage. Thus the computerbase results are left and while AMD are better, it isn't that much. 14nm era Intel was quite a bit behind 7nm Ryzens. The gap has closed a lot since then. Most of the peak power complaints against Intel could be largely resolved through power limit setting like AMD does.
 
I wasn't (still?) not sure what you're aiming for so I mentioned that earlier with the new products in mind.
Yeah. I'm not clear myself on what I want. Makes it tough for you guys to give recomendations when I keep moving the bar. Sorry.
 
AMD 5000 cpu failure rates where reported as high as 3%. Which isn't great in my opinion. Yields mostly to expect under 2% for any cpu manufactured.

AMD while running 2 CCDs take a memory writes penalty pretty hard as well, something that may not matter to gamers, but would reflect with productivity.

12400F core for core is just faster clock for clock than AMD 5000 series chips.

Knowing that, you'd buy into AMD 7000 series and not consider AM4 chips at this time.

In my opinion of course.
 
With Intel CPUs, if you're looking previous-gen and mid-range, they're competitive (right @DaveB ?).
IMO yes, 12th and 13th gen especially because you can go with your existing DDR4, for Zen 4 AMD you have to go for DDR5. I rode AMD from Zen, to Zen + through Zen 2 all the way up to a 12C/24T Ryzen 7 3900X. I just got tired of the AGESA and other annoying issues with each new gen. I've been running my 12C/20T i7-12700F/Asus H610 combo since January with zero issues of any kind. But I'll give AMD a shot again sometime, I always do.
 
OK guys. I'm in the market for a new gaming rig. Nothing fancy but building around a 3080Ti. I'm probably an AMD guy but I don't understand Intel CPUs and I'm looking for someone who knows better. Simple stuff here. I'm not looking to break the knowledge bank here.

P cores and E cores. What gives? Performance and economy? If a CPU has 6 P cores and 4 E cores, it gets listed as 10 cores. If they are different, how do I compare the performance of a "10 core" CPU to AMD or another Intel?

In terms of actual recomendations (if you have one) I mostly do Folding @ Home. Gaming is limited to Total War: Shogun 2, Civ VI and other lower tier games but FPS standards. I Fold, game a little game, stream YouTube mostly, email and web browsing.
Your a Waffler :rofl: (Waffler: One who can't make up their mind or flips back and forth))
 
AMD 5000 cpu failure rates where reported as high as 3%. Which isn't great in my opinion. Yields mostly to expect under 2% for any cpu manufactured.

AMD while running 2 CCDs take a memory writes penalty pretty hard as well, something that may not matter to gamers, but would reflect with productivity.

12400F core for core is just faster clock for clock than AMD 5000 series chips.

Knowing that, you'd buy into AMD 7000 series and not consider AM4 chips at this time.

In my opinion of course.
I'm not sure how high is the real DOA or dead after a short time of usage. I know there was a problem with the memory controller, and AMD was replacing every CPU that couldn't even boot above DDR4-3200. I had 2 chips like that.

I can't really see a lower memory write performance in AIDA64 tests. Actually, more threads give a higher bandwidth in these tests. Maybe I'm missing a specific workload, but at least in multithreaded tests, it looks like this. The same in Ryzen 7000 series, you get about 10GB/s higher bandwidth on 7900X/7950X than on 7600X.

In general, I see it like AMD just made something new, and there are no expected new product problems, which is quite surprising. Motherboards are ready out of the box and usually don't need BIOS updates. Intel is just pushing "old" technology to the limits. Lower chips are fine and have quite good results at lower wattage, but higher chips are running at the edge of throttling and require very good coolers. I feel that the highest chips are designed only to beat AMD at all cost. This is why they added many more E-cores to some chips = bumped results in multithreaded tests, while most games are not scaling well above 4-6 threads, so they can boost high and show good results on P-cores.
I'm not saying that any option is bad. All new CPUs are more than users actually need. It's just that Intel feels like is trying too hard to beat AMD recently.
 
AMD while running 2 CCDs take a memory writes penalty pretty hard as well, something that may not matter to gamers, but would reflect with productivity.
I thought it was the other way around. Since CCD writes are half speed compared to reads, you need 2 CCDs to max out writes.

On that note, I don't understand Zen 4 based on AMD comments so far and how they compare to Aida results. They said IF bandwidth was same (per nominal clock), by having links half as wide but twice the speed compared to Zen 3. Single CCD models are achieving more write than should be possible, so I have an understanding gap somewhere.

In general, I see it like AMD just made something new, and there are no expected new product problems, which is quite surprising. Motherboards are ready out of the box and usually don't need BIOS updates. Intel is just pushing "old" technology to the limits.
Zen 4 is a more radical change over Zen 3 for sure. AVX-512 addition is welcome by me and makes it much more attractive. Raptor Lake is not an architecture change over Alder Lake, but the process update they gave it does seem to have bought some good efficiency improvement. Worth at least a + under the old name scheme. It was reported that Raptor wasn't originally planned, but added in as Meteor is a bit late. Intel is still working on their process problems. If I ignore AVX-512, I do wonder if the i5 13600k would be as much CPU as I really need, and it makes lower Zen 4 offerings seem very expensive in comparison.
 
Back