14900 KS - the fastest Intel processor ever just launched

[Dr. Dro]

I'm not saying that you're wrong, but some criticism actually has some ground. Latest gen i7s and i9s are extremely inefficient. Stating this obvious fact, while unnecessary, doesn't constitute AMD fanboyism.

I don't think they're inefficient (again, care must be taken not to conflate efficiency and economy), but Intel does ship them beyond their optimal v/f curve points. AMD's been doing that too, on both sides you can shed up to 40% power by giving up on like 5% of performance.

 

[AusWolf]

I don't think they're inefficient

It's not an opinion. It's fact:

on both sides you can shed up to 40% power by giving up on like 5% of performance.

Sure, except that with a 7800X3D, you don't have to.

 

[Dr. Dro]

It's not an opinion. It's fact:
View attachment 339231View attachment 339232


Sure, except that with a 7800X3D, you don't have to.

By efficiency I don't meant perf/W, I mean architecturally-wise. Raptor goes brrr at the cost of going brrr in power too, Zen 4 does that to a much lower degree because it can't cross into the high 5-low 6 GHz domain. And yes, 7800X3D is probably perf/W (aka it's an economic CPU) and gaming king, great CPU imo, although lacking a bit in brute performance that the Core processor provides

 

[dgianstefani]

It's not an opinion. It's fact:
View attachment 339231View attachment 339232


Sure, except that with a 7800X3D, you don't have to.

It's not surprising to see K series chips tuned for performance not power draw. That's the target audience. Now look at non K chips, or even T chips.

 

[Jacky_BEL]

Temperature is the manifestation of heat and heattransfer.

Energy efficient processes are key to minimise heatloss and therefor low operating temperatures.

It is crazy to see GPU's and now CPU's draw so much power. I believe they are working deep in the area of diminishing returns , and thus not efficient.

 

[Gica]

Why complain about how inefficient it is if you are an AMD user with no intention of ever buying Intel? just pointless bleating.

When both efficiency and performance were in Intel's camp, they attacked the prices. In the period 2017-2020, the multicore performance was a reason for praise for AMD owners, all of them were rendering and encoding. Now, when an i5 fights with an r7 in multicore, their "special operation" has moved to consumption and that 1FPS won in games with 30% extra price, all "just gaming".
They will always find a reason.

 

[AusWolf]

By efficiency I don't meant perf/W, I mean architecturally-wise.

What the heck are you talking about? Efficiency is perf/W.

And yes, 7800X3D is probably perf/W (aka it's an economic CPU) and gaming king, great CPU imo, although lacking a bit in brute performance that the Core processor provides

If you need that extra brute performance for some application, then go ahead, buy an i7 or i9, there's nothing wrong with that.

It's not surprising to see K series chips tuned for performance not power draw. That's the target audience. Now look at non K chips, or even T chips.

I have a non-K 11700 that can draw up to 180 Watts in a PL-unlocked state.

The only reason why it sips power at stock is because it's limited to 65 W by default, which also means that it's limited to 2.6-2.8 GHz all-core, with R5 3600-like performance.

I imagine, newer non-K CPUs are similar in this regard: they are the same power-hungry beasts as the K variant, with a much more aggressive default power limit.

 

[dgianstefani]

What the heck are you talking about? Efficiency is perf/W.


If you need that extra brute performance for some application, then go ahead, buy an i7 or i9, there's nothing wrong with that.


I have a non-K 11700 that can draw up to 180 Watts in a PL-unlocked state.

The only reason why it sips power at stock is because it's limited to 65 W, which also means that it's limited to 2.6-2.8 GHz all-core, with a R5 3600-like performance.

We're not talking about backported to 14 nm Rocket Lake though.

 

[AusWolf]

We're not talking about backported to 14 nm Rocket Lake though.

Sure, but the way they achieve better efficiency is the same: with a much more aggressive default power limit. Once you disable it, you get 99% of the K version's power and performance.

 

[dgianstefani]

What's the point here?

Yes, changing settings in the BIOS changes efficiency. The same thing applies to K series if you want to go down that route.

 

[FoulOnWhite]

I'm not saying that you're wrong, but some criticism actually has some ground. Latest gen i7s and i9s are extremely inefficient. Stating this obvious fact, while unnecessary, doesn't constitute AMD fanboyism.

It has, but when every thread re intel gets posts that are it seems to me just an excuse to bash them, why post at all?

 

[kapone32]

Hot = observation of temperature e.g. having a high degree of temperature.

Heat = the quality of being hot.

What you're talking about is a physics explanation where the correct term would be heat flow or heat transfer.

I used the correct terminology.

Heat may be "correlated" with power draw, but it's more closely correlated with temperature resistance gradients, e.g. how easily "heat" transfers away from its source and diffuses into the environment. Incidentally, this is why cooling in space is very difficult, no medium to dump heat into.

A stock 7800X3D is "hotter" than a stock 14900KS, because the temperature resistance gradient is worse, despite a much higher "efficiency", or lower power consumption.

The reason why the 14900KS tested "hot" in TPU testing is because we used a upper mid range air cooler, so the limit was the cooling capacity not the chip. Put both a 7800X3D and a 14900KS under a 360 mm AIO or custom loop and the 14900KS will run "cooler".

You are of course welcome to check my definitions.

Why are you comparing this to a chip that would cost less even when you add a MB and RAM? I seriously doubt that a 360 AIO will cool a 7800X3D worse than a 14900KS. The rules of Thermal Dynamics cannot agree with the notion that a part producing 65W (90 if you want to strectch it to the M2 and RGB as well) will be hotter than a 400W part in the same space. You cannot spin this into a win for this chip. Intel does not even use a stock backplate either to absorb some of that heat.

 

[AusWolf]

What's the point here?

Yes, changing settings in the BIOS changes efficiency. The same thing applies to K series if you want to go down that route.

My point is that lowering the power target (for example, on a non-K, or T chip) lowers performance as well, which doesn't help overall efficiency much.

It has, but when every thread re intel gets posts that are it seems to me just an excuse to bash them, why post at all?

Sure, stating an obvious fact is unnecessary. It doesn't mean we can't have an educated conversation about it, though.

 

[dgianstefani]

Why are you comparing this to a chip that would cost less even when you add a MB and RAM? I seriously doubt that a 360 AIO will cool a 7800X3D worse than a 14900KS. The rules of Thermal Dynamics cannot agree with the notion that a part producing 65W (90 if you want to strectch it to the M2 and RGB as well) will be hotter than a 400W part in the same space. You cannot spin this into a win for this chip. Intel does not even use a stock backplate either to absorb some of that heat.

You can argue as much as you want but the temperature readings don't lie.

It's the AMD fanboys who bring up the 7800X3D every time an Intel CPU is reviewed, conveniently because it's the only AMD part that comes close to being more efficient in both single and multi threaded than an Intel equivalent. Besides, non K series parts aren't on this chart, but you can emulate them by looking at the power limited 14900K.

 

[FoulOnWhite]

My point is that lowering the power target (for example, on a non-K, or T chip) lowers performance as well, which doesn't help overall efficiency much.


Sure, stating an obvious fact is unnecessary. It doesn't mean we can't have an educated conversation about it, though.

Absolutley nothing wrong with good discussion, it's product bias i don't like.

 

[dgianstefani]

My point is that lowering the power target (for example, on a non-K, or T chip) lowers performance as well, which doesn't help overall efficiency much.


Sure, stating an obvious fact is unnecessary. It doesn't mean we can't have an educated conversation about it, though.

Except the lowered power consumption still leads to better efficiency, right down to 35 W as the chart I just posted shows. I'm not talking about Rocket Lake as I consider that irrelevant and non representative due to being on a node it wasn't designed for.

Plus, this thread is about Raptor Lake.

 

[AusWolf]

You can argue as much as you want but the temperature readings don't lie.

It's the AMD fanboys who bring up the 7800X3D every time an Intel CPU is reviewed, conveniently because it's the only AMD part that comes close to being more efficient in both single and multi threaded than an Intel equivalent. Besides, non K series parts aren't on this chart, but you can emulate them by looking at the power limited 14900K.

View attachment 339247View attachment 339248

By power limiting an i7 or i9, you gain a bit on the MT efficiency, but you lose on gaming performance. Limiting a 14900K to 65 W, which is around the maximum level the 7800X3D consumes in gaming, puts it on par with a R5 7600. Therefore, gaming = 7800X3D, MT performance = 14900(K). Saying anything about the 7800X3D in MT apps is just as much a waste of effort as claiming gaming efficiency on a 14900K.

 

[dgianstefani]

By power limiting an i7 or i9, you gain a bit on the MT efficiency, but you lose on gaming performance. Limiting a 14900K to 65 W, which is around the maximum level the 7800X3D consumes in gaming, puts it on par with a R5 7600. Therefore, gaming = 7800X3D, MT performance = 14900(K). Saying anything about the 7800X3D in MT apps is just as much a waste of effort as claiming gaming efficiency on a 14900K.

View attachment 339250

Well I didn't bring up the 7800X3D, kapone did. Look at the chart and decide for yourself how efficient these processors are.

It's perfectly possible to get good gaming performance when reducing power limits. You just have to set the voltage/frequency yourself and do a per core tune. Power limiting alone runs off the stock voltage frequency curve so it's not surprising lowering it reduces single core clocks.

 

[AusWolf]

Except the lowered power consumption still leads to better efficiency, right down to 35 W as the chart I just posted shows. I'm not talking about Rocket Lake as I consider that irrelevant and non representative due to being on a node it wasn't designed for.

Plus, this thread is about Raptor Lake.

In MT, yes. In gaming, no. The 14900K is already behind in gaming with a much higher power consumption. Power limiting it puts it even more behind.

Well I didn't bring up the 7800X3D, kapone did.

Fair enough. Personally, I think any 7800X3D vs 14900K(S) discussion is the equivalent of an apples to oranges comparison. They're entirely different products made for entirely different purposes.

 

[dgianstefani]

In MT, yes. In gaming, no. The 14900K is already behind in gaming with a much higher power consumption. Power limiting it puts it even more behind.

It's "behind" in gaming when paired with RAM 2000 MT slower than it can run, at 1080p with a 4090 in singleplayer games.

TPU tests both platforms at 6000 MT to standardize test rigs for control, but that's really only the AMD sweetspot.

Let me put it another way. There's good reason practically every streamer and esports pro uses Intel CPUs and NVIDIA GPUs in their multiplayer games.

 

[AusWolf]

It's "behind" in gaming when paired with RAM 2000 MT slower than it can run, at 1080p with a 4090 in singleplayer games.

TPU tests both platforms at 6000 MT to standardize test rigs for control, but that's really only the AMD sweetspot.

Let me put it another way. There's good reason practically every streamer and esports pro uses Intel CPUs and NVIDIA GPUs in their multiplayer games.

Sure, you can spend 50% more on a CPU that eats multitudes more power, and 50% more on faster RAM to achieve a near imperceptible performance advantage in a few select games, but why would you?

 

[dgianstefani]

Sure, you can spend 50% more on a CPU that eats multitudes more power, and 50% more on faster RAM to achieve a near imperceptible performance advantage in a few select games, but why would you?

Let's not keep moving the goal posts.

You know very well why someone would spend that extra.

To get the fastest processor, and not just for gaming.

 

[AusWolf]

Let's not keep moving the goal posts.

You know very well why someone would spend that extra.

To get the fastest processor, and not just for gaming.

So what's the goalpost, then? The i9 is good at raw MT performance, the R7 is good at gaming efficiency. I never doubted any of this.

Equally, I also believe that the i9 is not a gaming CPU because of its price inefficiency and performance inefficiency, just as much as the R7 is not a productivity one because of its relatively low MT performance. I don't know why some other members keep arguing otherwise in every thread.

 

[dgianstefani]

So what's the goalpost, then?

The i9 is good at raw MT performance, the R7 is good at gaming efficiency. I never doubted any of this.

Equally, I also believe that the i9 is not a gaming CPU, just as much as the R7 is not a productivity one. I don't know why some other members keep arguing otherwise in every thread.

Because it is. Just like it is a good productivity CPU. If you want to claim otherwise that's your right, but this is an open forum and people will contest you on that.

 

[Gica]

mac mac mac and oac oac oac

The height of embarrassment: to discuss the lack of efficiency of the processors you don't own, but not to criticize the efficiency of the video card you bought.

In 1440p gaming, according to the TPU review, the 7800XT beats the 4070 by only 3.7% with a 19.7% higher power consumption.

It's incredible how low you can bring the discussion. Invade Intel topics with the same impudence, although it is clear that you do not have one of the same class (alder-raptor), and some have not owned Intel in their lives, they just read other people's stories.