I bought the 7600 for the reasons Humbug mentions, it performed close enough to the more expensive options in gaming that it was good enough for now at 1440p.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD Zen 5 rumours
- Thread starter Grim5
- Start date
Caporegime
So why you buy half a CPU then last time round ?TBH i would, but i play games that are thread heavy.
Infact i did..... last gen tho
Caporegime
Soldato
Real world testing is much different then what reviews and benchmarks show you..I've done a lot of testing with many CPUs, and I can tell you that for example, turning off the ecores on my 12900k and only using those 8 physical P cores makes lots of games much much worse. I can only imagine how much worse it would be with just 6. From loading times to responsiveness to actual framerate in heavier scenes and areas,, you need cores plain and simple. If you are into the latest triple A games don't settle for 6 cores, there is a big difference.
Some work types run well on a mesh topology and the tech scales reasonably well without too much of a latency penalty. Some workloads don’t run well on a mesh though. The ringbus side of Intel chips (those are called the “P” core, Bencher) don’t scale well at all with core count. If Intel keeps increasing per core performance core count might have to decrease.
When talking of more cores on Intel you have to define which ones you would like more of. Sandybridge or Atom?
Last edited:
Caporegime
Try that with games that saturate the 16 P-Core threads and spill over in to E-Cores not by choice but needs.
Very few games will do that but i know of one and despite its developers doing their damnedest to get it to run smoothly on these CPU's it just wont.
That's a potential problem further down the road as more games become like that. Its one reason i talked about this yesterday.
For gaming, when it boils down to it they are 8 core CPU's.
PS: by saturate i don't mean fully load, just efficient thread balancing.
Last edited:
I don't want to take sides but... Total cost of platform right now makes the i7-13700k with DDR4 pretty much the best option VS anything AMD, meaning I can either take 8 AMD cores or 8+x Intel cores at a similar price and performance.
Caporegime
If you're going D4 and 16 threads my preferred choice would be a 5800X3D.
But 13700k is 24 threads and faster than 5900x at MT and at least as fast as 5800X3D in gaming, that's the point.
Last edited:
Caporegime
You quoted the post where i talked about the potential for problems with these hybrid chips.
My reasoning is something simple that doesn't have the potential for headaches in subsequent years. Its just as quick in games as the 13700K, the chip, a decent board and 32GB of ram all comes to little over £500, that's the whole damned platform and no need to worry about E-Cores messing with your gaming performance, when talking about my own perspective i already wouldn't buy one of these chips full stop.
I realise its specific to me and others like me, a lot of who have these chips and do nothing but cry about not being able to get stable gameplay unless they turn those half cores off, but that's what it is.
I take nothing away from these chips, i mean it when i say i think they are good, i hold to that, but ultimately i see them as potentially compromised.
Last edited:
Yes, I know what you're saying and I'm not disagreeing.
The problem is: Will we get bigger performances issues from the problems you say or from being thread limited?
Let's take it to a slightly different field:
Will a 7800X3D perform better than a 13700k in 5 years if games become more multi-threaded?
Caporegime
Ok, TBH i don't want to make hard predictions, which is why i qualified what i said as a potential, there is no way of knowing for sure.
You might be right, the E-Cores might take the pressure off the main cores when they get overloaded.
What i'm saying in that rearguard, from (the one example) of that which i have seen so far.... its better just to switch the E-Cores off, probably because you just get a slowdown in draw calls, a CPU bottleneck rather than an uneven and erratic draw call latency caused buy uneven performance cores. which is what causes the stuttering and frame hanging.
Last edited:
Caporegime
Most games, by most i mean all so far as i can tell, run on 32Bit Math. That gives you about a few dozen KM of simulated 3D world squared, so square KM of simulated world numbering in the thousands.
Ok, so Star Citizen currently is 70,000,000 KM squared, 4 planets, 12 moons, 4 halo cities that on their own are too larch to fit inside 32Bit math, a dozen space stations inside volumetric gas clouds big enough to swallow the earth 10X over, hundreds of physicalist ships each with dozens of physicalist components all owned by 100+ players each with hundreds of physicalized inventory items and tens of thousands of persistent empty drink bottles people littler floors with all moving about inside the gravity of planets and moons moving about in day / night cycles around a star.
Just to give you a vague perspective, that needs to run in 64Bit math, on your CPU, when CIG said this is what they were going to do industry veterans laughed "your ambition is impossible"
Its a amazing it runs at all without a Nasa super computer to simulate a solar system and its contents. and your CPU knows it.
Ok, so Star Citizen currently is 70,000,000 KM squared, 4 planets, 12 moons, 4 halo cities that on their own are too larch to fit inside 32Bit math, a dozen space stations inside volumetric gas clouds big enough to swallow the earth 10X over, hundreds of physicalist ships each with dozens of physicalist components all owned by 100+ players each with hundreds of physicalized inventory items and tens of thousands of persistent empty drink bottles people littler floors with all moving about inside the gravity of planets and moons moving about in day / night cycles around a star.
Just to give you a vague perspective, that needs to run in 64Bit math, on your CPU, when CIG said this is what they were going to do industry veterans laughed "your ambition is impossible"
Its a amazing it runs at all without a Nasa super computer to simulate a solar system and its contents. and your CPU knows it.
Last edited:
I think they laughed at the unnecessary self imposed hurdle of having it all in same coordinate system
why not have space scale things in space scale coordinates and human scale things in more precise coordinates? Its not like we need to measure exact distance from corner of this table to surface of nearby planet all the time?
even better to keep these separate, could run calculations in separate threads.
not dismissing need for fast 64bit maths, just this particular example
Last edited:
Caporegime
Its necessary if what you're trying to is make a sandbox universe without limits, for example ships with full interiors fully physicalized with physicalized components. This is important for the true (Star Wars) multicrew experience because if you break the game world up into load screens you can't have an independent crew in that ship, what happens to them when you load from one instance in to another? Well you just can't do it, so no true multicrew experience, this is why all the other up coming "Star ####" copy cat games (i don't mean that derogatory) and similar existing games don't have that.
Just one of many examples.
Last edited:
Soldato
Lots of games need / use the ecores. Cyberpunk, warzone, tlou, spiderman, they all perform much more smoothly with them on than off. Ecores are used as a substitute for HT, whenever a game was going to load HT because the physical cores weren't enough, if it finds ecores it uses those instead.
If a game performs worse with ecores on, especially on 13th gen, something is really wrong with the game. It probably bypasses the built in hardware scheduler and tries to do stuff the stupid way.
Caporegime
You make a compelling point on the HT threads.
Soldato
Ikr, HT is obnoxious for games. That's why I've been begging for a 12 core CCD 3d, to turn of that pesky HT and enjoy 12c low latency 3d cache goodness. But until then, intel is the way for me
Last edited:
Probably. Now we have to get into cache sizes and the limitations of the Atom cores. The Atom cores are a help but also a hindrance.