Intel 18 core: disappointing early indications.

Ready4Dis

2[H]4U
Joined
Nov 4, 2015
Messages
2,523
https://www.techspot.com/news/81496-intel-10th-gen-18-core-beats-core-i9.html

This is a preview of the 18/36... Losing to the 3950x easily in both single and multi threading. Given Intel's track record on pricing, I'm sure it will cost at least twice as much. I wouldn't be surprised if it burn d more power to do so either. Intel needs to work out it's 10nm process soon because their armor is starting to show some signs. Can't wait to see the real numbers in other tasks, but it looks like it's going to go AMDs way in most tasks and pricing should by an easy crown to wear unless Intel does something spectacular with pricing (which they can't without competing with themselves).

Be interesting to see how these actually perform in real world testing.
 
AMD has a process advantage (it feels weird saying that) as well as a technology advantage at this point. Intel is going to have a hard time catching up on its current process node while trying to stay in TDP budgets against anything AMD has to offer. And then there's price too...
 
It's an engineering sample running at 2.2/3.2ghz. I'm sure it won't run at those speeds in production samples.

If the 3900x gets hot, Intel is going to have an even harder time managing heat. It may not clock much higher without exotic cooling.
 
It's an engineering sample running at 2.2/3.2ghz. I'm sure it won't run at those speeds in production samples.

Well that is sort of the problem with their 10nm process. I'm sure they will get a bit more... but they aren't going to get a miracle 50% clock bump or something.

The best Intel can hope for on this part is a few selected benchmarks getting close. They might even be able to show a few wins. Problem is Intel isn't going to have more then a handful to ship and they are going to charge 20-50% more min. AMDs 3950 will ship in volume. We know they are turning out fully functioning 8 core chiplets with zero issues based on supply of 3700/3800.

AMD will have the likely faster part... at a cheaper price... in stock ready to purchase. There is a reason Intel has accelerated their 7nm node switch. 10nm is a lost cause.
 
  • Like
Reactions: B770
like this
It's an engineering sample running at 2.2/3.2ghz. I'm sure it won't run at those speeds in production samples.
This is supposed to be Intel's answer to the 3950x... If they are still at the point of engineering samples that can't clock, it's even further behind than I thought. Just think, this is beating Intel's own 9980XE, which is probably their target, by 10+% in mutlithreading, which isn't to bad from on generation to the next... It's just AMDs offering is looking closer to 25% faster than the 9980XE at about 1/2 the price, so unless they were running it slow just to check things out and are going to ramp up speeds without requiring a cooling loop, it's going to be a hard sell.
 
This is supposed to be Intel's answer to the 3950x... If they are still at the point of engineering samples that can't clock, it's even further behind than I thought. Just think, this is beating Intel's own 9980XE, which is probably their target, by 10+% in mutlithreading, which isn't to bad from on generation to the next... It's just AMDs offering is looking closer to 25% faster than the 9980XE at about 1/2 the price, so unless they were running it slow just to check things out and are going to ramp up speeds without requiring a cooling loop, it's going to be a hard sell.

It doesn't prove that at all. Engineering samples are almost always clocked significantly lower than production parts.
 
I suspect Intel will have something competitive come October, question is can they deliver on it? Having the better part is great but if you can't get them to market in any meaningful quantity then that doesn't do anybody any good.


*Mumbles to myself "I'm not butt-hurt over my 3900x still not having shipped yet or anything."
 
Anything that forces Intel to reevaluate their pricing model is good for everybody.

Personally I care more about chipset which, unfortunately, is AMD's weakest link. Woe be to Intel should AMD ever develop a rock solid desktop/HEDT platform.
 
Anything that forces Intel to reevaluate their pricing model is good for everybody.

Personally I care more about chipset which, unfortunately, is AMD's weakest link. Woe be to Intel should AMD ever develop a rock solid desktop/HEDT platform.

I have seen several issue with ASUS's X370 boards (Crosshair anyone?)
However with my ASrock board, never had any problems with it other than not being able to run my 3000mhz ram @3000Mhz at first, Ran it at 2666 and it was stable. AMD released AGESA for faster RAM supprt and set it to 2933 and its been running like that ever since!
I'm a ASrock believer lol!
 
Will be interesting to see how this plays out. AMD crushes Intel in high thread benchmarks. This had me regretting my decision to go Intel on my build. After some further research, it turns out that AMD benchmarks well but falls flat when it comes to real workloads.

The problem with AMD’s chiplet approach is that the performance gets bottlenecked whenever the task has to utilize a resource outside of the chiplet. Synthetic benchmarks are fairly trivial tasks repeated quickly, so the chiplets score well there. Real media-laden workloads (think 50+GB RAM) end up cutting performance in half on AMD.

If AMD can get real workloads up to snuff, they’re well positioned to put a world of hurt on one of Intel’s highest margin segments.
 
Motherboards make a big difference for the people who want somehing dependable AMD boards usually have better reviews on Newegg.
 
Will be interesting to see how this plays out. AMD crushes Intel in high thread benchmarks. This had me regretting my decision to go Intel on my build. After some further research, it turns out that AMD benchmarks well but falls flat when it comes to real workloads.

The problem with AMD’s chiplet approach is that the performance gets bottlenecked whenever the task has to utilize a resource outside of the chiplet. Synthetic benchmarks are fairly trivial tasks repeated quickly, so the chiplets score well there. Real media-laden workloads (think 50+GB RAM) end up cutting performance in half on AMD.

If AMD can get real workloads up to snuff, they’re well positioned to put a world of hurt on one of Intel’s highest margin segments.

Have you even looked at the AMD Rome benchmarks..........................
Plenty of results that use 50+ GB of RAM and it utterly decimates Intels fastest most expensive Xeons.................................
 
Will be interesting to see how this plays out. AMD crushes Intel in high thread benchmarks. This had me regretting my decision to go Intel on my build. After some further research, it turns out that AMD benchmarks well but falls flat when it comes to real workloads.

The problem with AMD’s chiplet approach is that the performance gets bottlenecked whenever the task has to utilize a resource outside of the chiplet. Synthetic benchmarks are fairly trivial tasks repeated quickly, so the chiplets score well there. Real media-laden workloads (think 50+GB RAM) end up cutting performance in half on AMD.

If AMD can get real workloads up to snuff, they’re well positioned to put a world of hurt on one of Intel’s highest margin segments.
That concept would be a first.
 
Will be interesting to see how this plays out. AMD crushes Intel in high thread benchmarks. This had me regretting my decision to go Intel on my build. After some further research, it turns out that AMD benchmarks well but falls flat when it comes to real workloads.

The problem with AMD’s chiplet approach is that the performance gets bottlenecked whenever the task has to utilize a resource outside of the chiplet. Synthetic benchmarks are fairly trivial tasks repeated quickly, so the chiplets score well there. Real media-laden workloads (think 50+GB RAM) end up cutting performance in half on AMD.

If AMD can get real workloads up to snuff, they’re well positioned to put a world of hurt on one of Intel’s highest margin segments.

*waits patiently for examples of such behavior*
 
Memories. Back in the Opteron hayday. Before AMD acquired ATI.

One of the best chipsets came from Nvidia. If you wanted stability and full features, in the datacenter, you chose the Nivida chipset.

Then, AMD bought ATI (instead of Nvidia) and the rest is history...
 
  • Like
Reactions: N4CR
like this
It doesn't prove that at all. Engineering samples are almost always clocked significantly lower than production parts.

Not necessarily. It depends on at what stage the engineering sample was sent out. Earlier silicon will typically have reduced clock speeds while later engineering samples leading up to the retail release are generally identical to retail silicon regarding clock speeds.
 
It doesn't prove that at all. Engineering samples are almost always clocked significantly lower than production parts.
No, it's not proof, they could surprise and get an all core of 5ghz... Or somewhere in between. Right now these are the only numbers we have to go by. Hence the "early indications" in the title. I'm hopeful they can at least get the frequency up a bit without hitting 200+ tdp, but it's a big unknown. Like I said, I'm actually hopeful this isn't their last #'s, but time will tell if course.

PS. I couldn't find any info on when this is to be released, if it's a long way out it makes sense for the low numbers, but if it's to compete with the 3950x... I would hope it's not a year out from release.
 
It's an engineering sample running at 2.2/3.2ghz. I'm sure it won't run at those speeds in production samples.

What speeds do you expect? It kinda defeats the purpose of an engineering sample if it doesn't replicate the physical requirements of the production chip, and if it was just some dud sample from a bad first production run it never would have been seen outside of Intel.

The motherboard makers are the ones who leak ES chips, if that's the case here I'm guessing this is pretty close to the final product.
 
I think we should expect speeds similar to the 9980XE, +/- a few hundred MHz depending on how they're planning to optimize the design. They can't realistically drive the power consumption significantly higher and still actually plan on selling them to consumers (not that this has stopped them before), so they can't squeeze that much more clockspeed out of the chips than they get now.
 
Will be interesting to see how this plays out. AMD crushes Intel in high thread benchmarks. This had me regretting my decision to go Intel on my build. After some further research, it turns out that AMD benchmarks well but falls flat when it comes to real workloads.

The problem with AMD’s chiplet approach is that the performance gets bottlenecked whenever the task has to utilize a resource outside of the chiplet. Synthetic benchmarks are fairly trivial tasks repeated quickly, so the chiplets score well there. Real media-laden workloads (think 50+GB RAM) end up cutting performance in half on AMD.

If AMD can get real workloads up to snuff, they’re well positioned to put a world of hurt on one of Intel’s highest margin segments.


I want to see the receipts for these claims. Because the Rome benchmarks of many websites were real workloads and no such thing happened.
 
  • Like
Reactions: B770
like this
Interesting article and even more interesting comments > https://www.androidauthority.com/amd-vs-intel-994185/
Really odd timing for this piece... Less than a month before AMDs new parts dropped, then a per/watt section using TDP, which we all know aren't realistic. Not really sure what it was meant to show with that one. They picked the highest tdp per core as their reference... Why not the 3900x or 3950x that both have the same TDP as the 3800x? Seems like an odd data point. Maybe the 3700x that's got the same # of cores @ 65w? Really odd choice.
"Unfortunately, we don’t have benchmark numbers for comparison given the Ryzen 3000 desktop parts won’t arrive until July." Seems like they could've waited 3 weeks and had actual #'s. It's not like this was an article on strict/hard time limit like a new GPU release or, you know, the CPUs they're using in their comparison that was released soon.
Honestly it's a pretty good overview despite my critic's, but sure had some odd choices for some of the comparisons.
 
Yeah,
Will be interesting to see how this plays out. AMD crushes Intel in high thread benchmarks. This had me regretting my decision to go Intel on my build. After some further research, it turns out that AMD benchmarks well but falls flat when it comes to real workloads.

The problem with AMD’s chiplet approach is that the performance gets bottlenecked whenever the task has to utilize a resource outside of the chiplet. Synthetic benchmarks are fairly trivial tasks repeated quickly, so the chiplets score well there. Real media-laden workloads (think 50+GB RAM) end up cutting performance in half on AMD.

If AMD can get real workloads up to snuff, they’re well positioned to put a world of hurt on one of Intel’s highest margin segments.
I'm with the others....where have you seen this? Any links? Everything I have read says AMD scales really well, from their 4 core to their 32 core CPUs. Many threads is the place I see AMD pulling away from Intel in every review I've seen, single core and games Intel win most often.
 
It's very well established on the user forums for the various rendering products that 2990wx vs 9980XE ends up being a wash in real world usage even though the synthetic benchmarks tell a very different story. You can see a bit of this when you look at the built in benchmarks for Corona. On top of that, there's also viewport performance where TR2 typically gets completely smoked by Skylake.
 
It's very well established on the user forums for the various rendering products that 2990wx vs 9980XE ends up being a wash in real world usage even though the synthetic benchmarks tell a very different story. You can see a bit of this when you look at the built in benchmarks for Corona. On top of that, there's also viewport performance where TR2 typically gets completely smoked by Skylake.

You are a year behind.. please catch up and show us where this is true with the processors AMD has released this year, as well as Rome that is going to be released very, very soon along with the 3950x, which is what it is being compared to in the article in the OP. Nobody here is talking about AMD's performance from last year's processors.
 
You are a year behind.. please catch up and show us where this is true with the processors AMD has released this year, as well as Rome that is going to be released very, very soon along with the 3950x, which is what it is being compared to in the article in the OP. Nobody here is talking about AMD's performance from last year's processors.

Catch up? I'm not going to place wagers based on hopes and dreams for processors which haven't been released yet. People who do that are nothing beyond fanbois.

I can't believe you haven't even mentioned Intel's 7nm 28-core CPU which is going to be released in the future and is a huge step beyond a slightly earlier but also still unreleased AMD chip. likeomg.

What's funny is how desperately and quickly the fanbois came out to defend the honor of a company they like. You are all so desperate to carry that flag that you don't even realize you're actually agreeing with my original post.
 
It's very well established on the user forums for the various rendering products that 2990wx vs 9980XE ends up being a wash in real world usage even though the synthetic benchmarks tell a very different story. You can see a bit of this when you look at the built in benchmarks for Corona. On top of that, there's also viewport performance where TR2 typically gets completely smoked by Skylake.

"TR2 gets completely smoked by Skylake" and AMD's midrange parts are a whopping 5% slower than the 9900K in a benchmark that clearly doesn't scale beyond eight threads.

So basically what you're saying is you can't actually support your argument with evidence and now you're lashing out at those who called you out on your bullshit.

Good talk.
 
Viewport performance is a real world workload. Sorry for bringing reality into this.

And you're still agreeing with me. Hilarious.
 
Catch up? I'm not going to place wagers based on hopes and dreams for processors which haven't been released yet. People who do that are nothing beyond fanbois.

I can't believe you haven't even mentioned Intel's 7nm 28-core CPU which is going to be released in the future and is a huge step beyond a slightly earlier but also still unreleased AMD chip. likeomg.

What's funny is how desperately and quickly the fanbois came out to defend the honor of a company they like. You are all so desperate to carry that flag that you don't even realize you're actually agreeing with my original post.

WOW!! (LOL, you are in a thread arguing about a processor that isn't released yet, and then claim you are not going to place wagers on a processor that hasn't been released yet.... Do you see the failure in that statement?)

Now, show us something from the processors released this year by AMD. They have released 5 processors so far that you can show us examples that support your claim, if true. Not examples from processors released last year. I am ok with taking the 3950x or Rome out of the equation... So, show us.. we will wait for it, or you can admit that you claim above is false, because there is nothing out there that supports AMD falls flat on it's face as you claim where zen2 is concerned... Like i said . Please catch up to this year.
 
Last edited:
Now we are being called fanboys because you couldn't produce proof of such outlandish claims? Is this guy just trolling us or what?

I provided proof. The proof backed up my claim rather than yours. Now your chosen undergarments are wadded.
 
Viewport performance is a real world workload. Sorry for bringing reality into this.

And you're still agreeing with me. Hilarious.

Sure, we're agreeing that a benchmark that favors clockspeed over core count is won by the 9900K. That's not news and doesn't support your original assertion, which was, and I quote

The problem with AMD’s chiplet approach is that the performance gets bottlenecked whenever the task has to utilize a resource outside of the chiplet. Synthetic benchmarks are fairly trivial tasks repeated quickly, so the chiplets score well there. Real media-laden workloads (think 50+GB RAM) end up cutting performance in half on AMD.

So, again, please provide examples.
 
I provided proof. The proof backed up my claim rather than yours. Now your chosen undergarments are wadded.

If we where talking about last years technology, that would be true, but we are not talking about last years technology, we are talking about this years (those based on zen2). Once again, please show us proof tbat AMD falls flat where zen2 is concerned.

Even your "proof" shows the the 3800x as number 2 if you take out the 9900k overclock result, only behind the 9900k by only 61 points... So your own "proof" counters your own claim that AMD falls flat.
 
Last edited:
Sure, we're agreeing that a benchmark that favors clockspeed over core count is won by the 9900K

That's how the real world works, unfortunately. Workloads that are highly multithreaded are almost never exclusively multithreaded. This is why synthetic benchmarks often don't align with real world performance.

Dry up those tears though. There are still plenty of good reasons to go with AMD for these workloads (not having a meager 128GB RAM limit is a huge one). Historically, however, the real world performance gap has not been aligned with the synthetic benchmark gap. Like I said in my OP, if AMD has resolved that with the latest generation, then they will put a huge hurt on Intel in this market.
 
That's how the real world works, unfortunately. Workloads that are highly multithreaded are almost never exclusively multithreaded. This is why synthetic benchmarks often don't align with real world performance.

Dry up those tears though. There are still plenty of good reasons to go with AMD for these workloads (not having a meager 128GB RAM limit is a huge one). Historically, however, the real world performance gap has not been aligned with the synthetic benchmark gap. Like I said in my OP, if AMD has resolved that with the latest generation, then they will put a huge hurt on Intel in this market.

So a single test in which the 9900K beats the i9 9960X by the same margin as it does the Ryzen 5 3600 is conclusive proof that AMD's performance is bottlenecked by the use of chiplets? Also, the ability to use more than 128 GB of RAM is an advantage when, in your world, AMD has problems with "media-laden workloads" that use "50+GB RAM?"

You hear how stupid that sounds, right?
 
So a single test in which the 9900K beats the i9 9960X by the same margin as it does the Ryzen 5 3600 is conclusive proof that AMD's performance is bottlenecked by the use of chiplets? Also, the ability to use more than 128 GB of RAM is an advantage when, in your world, AMD has problems with "media-laden workloads" that use "50+GB RAM?"

You hear how stupid that sounds, right?

It probably does sound stupid to someone who doesn't run the workloads. No surprise there. This is [H], so real world performance is a joke and synthetics are what matter because that's what leaderboards show.
 
Thunderdolt could not be more aptly named...

Meanwhile, getting back on topic before we get yelled at by Kyle, anyone have any guesses as to where Intel could be getting the apparent IPC gains for this chip? The mystery chip would have to be nearly 30% better clock-for-clock if the reported specs are true.
 
Back
Top