Jump to content

Same old story as far as part optimization goes? Wildly different performance between CPUs?


Recommended Posts

Not only is the part count situation the same as the very beginning of KSP, the physics code as it relates to parts seems to be getting slower. Since taking a break from KSP and overclocking my computer, I am actually able to run less parts. The 700-part range was slow but doable for me, but now 300 parts is about the limit I can put up with.

My CPU is an AMD Phenom II hexacore overclocked to 4.0 ghz. I can see from gaming benchmarks that the latest Intel "Core" CPUs have about a 25% IPC (instructions per clock) advantage over AMD CPUs in gaming related tasks. The AMD Phenom II CPUs like I have are slighly faster in IPC (5%) than the latest AMD CPUs that can reach higher clock rates. Tell me, what then would account for a computer with an Intel "Core" CPU being able to run a ship with 1200 parts vs. my system only being able to run 300 parts? The instruction per clock difference is not enough to suggest such a staggering level of difference.

I have also thought of some things that are not the cause of the performance discrepancy to rule out any armchair computer "experts" giving BS answers:

Memory bandwidth/latency difference - all indications point to this not being a significant factor in game performance.

Storage device bandwidth/latency difference - all indications point to this not being a significant factor in game performance.

AMD vs. Nvidia - this literally makes no difference. Same bad FPS with a high end model of either brand of card.

Linux vs. Windows - for me the Linux version runs noticeably worse allowing about 100 fewer parts than the Windows version. This is despite Linux running much faster in general.

64-bit vs. 32 bit - no difference. 64-bit might even run slower when it's not crashing.

The only thing I can think that accounts for these problems are:

1. Low standards - all my ships must pass a rigorous flight test included extended level flights through the atmosphere at high speeds. Cobbled together ships may be able to reach orbit with more parts and less testing. If this is the case, I feel that misleading posts that show unusable ships should be removed from the forum to save people like myself the troubleshooting time trying to match standards that were never actually set in the first place because the person that posted the ship originally is in fact a horrible builder.

2. Intel CPU-related optimization in the Unity Engine Itself - SSEwhatever and whateverfeatureamddoesnothave are being used heavily thus leading to bad performance from AMD CPUs.

Someone save me from the boredom of small simple ships please! I want to join the huge-ships-with-good-fps master race.

Link to comment
Share on other sites

If this is the case, I feel that misleading posts that show unusable ships should be removed from the forum to save people like myself the troubleshooting time trying to match standards that were never actually set in the first place because the person that posted the ship originally is in fact a horrible builder.

What do you mean by "misleading posts" and "unusable ships?" Are you saying that you want people who build massive ships banned from the forums? Because that doesn't make any sense. Nobody is making you build massive 1000 part ships. Who cares if they want to make massively overbuilt and badly designed rockets. It doesn't affect you at all. If you don't want to build massive ships, it is entirely possible to go to Jool and back with a fifty part (or smaller) ship. And I doubt that KSP has slowed down to less than 50% of its old speed since whatever the last version you played on was. There have been numerous optimizations to the game engine over the years, so I suspect that the problem is on your end (maybe try adjusting the physics delta time).

Link to comment
Share on other sites

Do you have any hard evidence of comparative tests where a similarly rated Intel chip ran with 1200 parts and your AMD ran with 300? No?

Here's some: CPU Performance Database Thread

As for your two suggestions:

1. What? Part count is part count, I don't see how the design of the ship matters.

2. No. Unity does not utilize any platform-dependent optimizations.

Link to comment
Share on other sites

From my experience overclocking the Phenom II, focus on the northbridge. You'll improve overall system IO and get far more tangible results than just overclocking the CPU cores. 1600MHz is about right for 1333MHz DDR3. Somewhere on the tom's hardware forums there's a thread on how fast you need to go and what hypertransport multiplier to use given your build. I sold that rig years ago, so I honestly couldn't tell you how it plays KSP.

Link to comment
Share on other sites

There's not exactly a hard limit to part count, it varies greatly depending on the player's tolerance for lower framerates. If you build 1000+ part monsters, you are going to pay a performance penalty no matter the CPU you are using.

Removing pictures of high part count ships would be removing some of the most popular content here, I really don't think that's going to happen.

Link to comment
Share on other sites

This thread is quite old. Please consider starting a new thread rather than reviving this one.

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...