Impact on system performance

[Galatian]

When you say the best prospects in pentium are slower then core due we all agree except If you bring the arguement over to Civ series you see. For instance wheres your proof civ4 was programed to get FULL use of this upgraded tech for its benifit on INTERTURN, as opposed to ENHANCED mono optimization? .
You say all software is better run on core due but have no proof to invalidate the compaints coming from coredue players of civ4.
I bet you won't hear anyone complain with a rig the likes I linked (64bit mono 64 bit supported 2mb L2cache)
Ontop, its got muti thread advantages yet this means nothing and dual, sorry duo, does?

Please Im not workin in the computer biz IM sure you know your stuff, I know my civ stuff though, you be surprised how much I spent YOUTUBING this study. I want a benchmark link that proves this is infact the case. Otherwise, I go by what I can link here to contradict your claims(IN THE CIV DEPT ONLY!!!)..This is after all not a general tech discussion where generalization has any merit. This is civ specific and we need specific evidence not what is 'suppose' to happen in a majority of instances. Here I say we have article exception to the your 'truth'

Simply look at wikipedia or even just go onto the intel side. In fact the Mobile Pentium have been based on the Pentium 3 core and this is where intel realized that more Mhz doesn't mean more power...it only means a good advertisment. The Hyper Threading was invented espeically due to this problem. The pipelines of the Pentium 4 have been so large (so they can have a lot of clock speed), that they needed a better processing of the orders coming into the processor, which is what we know as Hyperthreading.
This technology would be absolutely inefficient on core 2 duos, because they have a different architecture with smaller pipelines. It's as simple as that ;-)

About the RAM issue: Your playing it on a norebook right? I'm using a 7900 Go GTS so it is slightly better then yours. I would imagine that your 7700 uses the system ram instead of it's own dedicated ram. So this means infact a huge limitation of ram: 1 GB almost for Vista (with Virus scanner ect. on) + 512 MB for the graphic card and then you know how much is left for BTS.

 

[Dano_01]

I have Q6600 @ 3.00Ghz, 4GB RAM, and 8800GT 512RAM and it's pretty slow at huge. It's unfortunate that the game is not optimized for multi-core CPUs. I am toward the end of tech tree and the wait is about 50 seconds per turn now.

I instant messaged a tech from Gateway today about my problem for around 35 mins. His best solution was to try a different game. Not very encouraging.

 

[Galatian]

Hmmm, Ithink they actually might be right. Try out a different game and check how that is working. Your System sure doesn't sound bad, which is why we need to figure out whether it is a problem with something on your system or if it is a problem related to BTS (which I tend to believe considering the countless threads about performance issues since BTS)

 

[Thanny]

Definitely find out about the version of Windows; is it the 32-bit edition or the 64-bit edition? So long as you're using the 32-bit edition, 4gb of RAM isn't doing you any good, in fact it can actually slow you down. 32-bit OSes don't really handle more than ~3gb of RAM (thus why 64-bit editions are being phased in). Check this out first.

32-bit Windows XP (and earlier) can only address 3.5GB of RAM (I have four, and that's how much it sees), but every other 32-bit OS I know of sees the full 4GB. I believe that includes 32-bit Vista as well.

In the case of XP, it does not slow anything down to have more - it just doesn't see it.

There have been memory addressing extensions in 32-bit processors for a while now, too, which an OS like Linux takes advantage of, meaning you can address 4GB and more in 32-bit Linux (total - still limited to 4GB per process).

For the graphics card, a quick test you could try is to run the game at a low monitor resolution and low graphics (game will look extremely ugly, but it's just a test). If the lag problems are still basically the same, then it's unlikely to be graphics lag, and the problem lies elsewhere. I don't honestly know whether that card should be able to handle rendering Civ4 at full resolution on a 24" widescreen or not.

Civ4 is a little weird in that department. I get 30+fps (more than enough for a turn-based strategy game) when I run with all high-quality graphics options on at 2560x1600 using my 8800GTX. At first. In late game, when there are thousands of units on the map, that slows down to a couple frames per second.

I did an experiment in one conquest-only game where I had almost won. With my couple thousand units, the game crawled while doing nothing during my turn. None of the graphics options made any significant impact, nor did halving the resolution to 1280x800. If I pressed the bare-map toggle, it zoomed back up in rendering rate - not back up to 30+, but smooth enough. If I deleted almost all my units, the same happened.

The only plausible theory I have at the moment to explain that is that each unit is calling out to Python, to allow customized animations, or something of the like.

 

[Thanny]

When you say the best prospects in pentium are slower then core due we all agree except If you bring the arguement over to Civ series you see. For instance wheres your proof civ4 was programed to get FULL use of this upgraded tech for its benifit on INTERTURN, as opposed to ENHANCED mono optimization? .

I'll just say, that's not how it works. The Core 2 processors execute x86 instructions more efficiently than the P4 processors (which isn't hard).

You're simply not going to find any single-threaded program that runs faster on a 3.8GHz Prescott than a 2.4GHz Core 2 Duo.

You say all software is better run on core due but have no proof to invalidate the compaints coming from coredue players of civ4.
I bet you won't hear anyone complain with a rig the likes I linked (64bit mono 64 bit supported 2mb L2cache)
Ontop, its got muti thread advantages yet this means nothing and dual, sorry duo, does?

Please Im not workin in the computer biz IM sure you know your stuff, I know my civ stuff though, you be surprised how much I spent YOUTUBING this study. I want a benchmark link that proves this is infact the case. Otherwise, I go by what I can link here to contradict your claims(IN THE CIV DEPT ONLY!!!)..This is after all not a general tech discussion where generalization has any merit. This is civ specific and we need specific evidence not what is 'suppose' to happen in a majority of instances. Here I say we have article exception to the your 'truth'

It seems you're saying I'm wrong about the clearly-expressed claim I make above. I'm not. It *is* a generalized answer, but it still applies to Civ4.

I'm not aware of what complaints about Civ4 you're referring to, except possibly that people are annoyed it doesn't actually use more than one work thread, so doesn't take full advantage of their CPU.

But the overall point, which is completely valid, is that Core 2 processor cores are faster than P4 processor cores, even with a more than 60% lower clock speed.

 

[Thanny]

In your case dualcore reverted back to faster monocore, aka pentium to take advantage of a game that dosn't utilize 2 more

Pentium 4 650 Prescott 3.4GHz 2MB L2 Cache LGA 775 84W Single-Core Processor

Since that's the model you chose, I'll show you some benchmarks comparing that to a E6600.

These are from the Tom's Hardware CPU charts, and are the few tests which I know are single-threaded. The two selected chips should show up in red:

Lame 3.98 Beta 3
AVG Anti-Virus 7.5.467
Unreal Tournament 2004
Xvid 1.1.2

As you can see, the Prescott chip is just outclassed. More clock cycles doesn't equal more performance.

 

[T.A JONES]

Since that's the model you chose, I'll show you some benchmarks comparing that to a E6600.

These are from the Tom's Hardware CPU charts, and are the few tests which I know are single-threaded. The two selected chips should show up in red:

Lame 3.98 Beta 3
AVG Anti-Virus 7.5.467
Unreal Tournament 2004
Xvid 1.1.2

As you can see, the Prescott chip is just outclassed. More clock cycles doesn't equal more performance.

I don't see anythin you linked thats relating to this discussion what-so-ever Oh wait, but up here I do,

I instant messaged a tech from Gateway today about my problem for around 35 mins. His best solution was to try a different game. Not very encouraging.

Civ4 is article exception. Take you general knowledge, generally speaking to the general 'compter forum' here on civfanatics. Here you provide facts that relate to the topic to back yourself up. Wheres Toms civ4 benchmark?

It possable your just guessing what ya expect would happen based on sole focus with these isolated and unrelated tests. ITs your blind eye to reports of what really has occured in the topic we discuss, that makes your knowledge on civ4's interturn crunching seem as inadequate as Core duo is when trying to run improvements on monocore 'groomed' software, like civ4 was

 

[Thanny]

It possable your just guessing what ya expect would happen based on sole focus with these isolated and unrelated tests. ITs your blind eye to reports of what really has occured in the topic we discuss, that makes your knowledge on civ4's interturn crunching seem as inadequate as Core duo is when trying to run improvements on monocore 'groomed' software, like civ4 was

Software simply does not work the way you sem to think it does. There's no such thing as optimization for a single-core chip. Civ4 does not magically run faster on older single-core Intel P4-based chips.

The benchmarks I linked to are not unrelated. They are of a variety of single-threaded programs run on the two processor types in question. They show definitively what you can expect from Civ4 regarding relative performance between the two different chips.

In fact, the chip you referred to - Prescott 650 at 3.4GHz - is slower than even the slowest Core 2 processor - the E4300 at 1.8GHz. That's for single-threaded programs, which includes Civ4.

For the record, I am a professional programmer. I'm not speculating about how these things work.

I realize you're feeling a strong inclination to defend the choice you've made, but the simple fact is that you're making a false claim. Civ4 does not run better on an older single-core processor than a modern dual-core processor.

 

[Calder]

You seem to be equating clock frequency with speed. The reason the Core 2 processors are such a big jump forward is not that they have two cores - Intel did that with P4-based chips, too. They're more efficient, executing more instructions per clock cycle.

The P4 was actually a giant step back in that respect - a P4 is much slower than a P3, Herz for Herz.

A single E6600 core at 2.4GHz is 20-40% faster than a single Prescott core at 3.8GHz, despite having a only a bit more than two thirds the clock speed.

To put it into more perspective, that same 3.8GHz Prescott single-core chip ranges from 5% faster (in a small number of things) to 20% slower than a 2.6GHz Athlon 4000+ single-core processor. Those are the fastest single-core offerings from Intel and AMD, and both are slower than a single 2.4GHz core from an E6600.

Good to see someone with a little basic knowledge which can really save you dollars when eying up a good gaming comp. It was 2 years ago when Civ IV just came out and I wanted to upgrade my old Windows 98 to something that will make Civ IV rock. I did a few months research on the internet and the result is now what I have is practically the norm for what you need even for today. And a little further hint, don't waste money on the GE Force 8 series cards just yet until the games you play are programmed to make use of DirectX10 by which time you're SLI & dual cored systems will certainly be useful.

 

[T.A JONES]

Software simply does not work the way you sem to think it does. There's no such thing as optimization for a single-core chip. Civ4 does not magically run faster on older single-core Intel P4-based chips.

The benchmarks I linked to are not unrelated. They are of a variety of single-threaded programs run on the two processor types in question. They show definitively what you can expect from Civ4 regarding relative performance between the two different chips.

In fact, the chip you referred to - Prescott 650 at 3.4GHz - is slower than even the slowest Core 2 processor - the E4300 at 1.8GHz. That's for single-threaded programs, which includes Civ4.

For the record, I am a professional programmer. I'm not speculating about how these things work.

I realize you're feeling a strong inclination to defend the choice you've made, but the simple fact is that you're making a false claim. Civ4 does not run better on an older single-core processor than a modern dual-core processor.

Nah, I don't defend the choice I made. I defend what I see is true in these circumstances . I don't dismiss fact based on observation over another generalized speel on how much better core due architecture should make civ4 over top model monocore. You think I havn't heard it before. Thing is the prrof wasn't there to back it up but plenty was their to dismiss this as the most logical guestimate

Saying your a programmer dosn't make it more true. Showing specific proof relating to your claim makes it so. Its a great big web and Civ4 is an award winning game number, #1 selling franchise!, surly they would have one measly core duo supremacy test to flex, No?

....unless, of course it was a small article exception to truth and of therfor lil consequence to coredue players who want to know they have best system in the world ITs a 'insignfigant truth' in the long run

Of course the Pent is dead, unless you just happen to be a civ player who chanced at the greatest mono perfomance for one third the price, then doubles it up nice with 64bit supported ram supply, 4.6ghz not breaking a sweat and plenty of L2 to make those turns fly.
You really think a 2.6 coreduo like Dano discribes is that much more efficient then a monocore blazing turns on mono designed software?

I know it hurts to hear an old computer beats you egostacks in some ways Heck, I heard the Intell marketing problem way back: Its convince buyers less is more, years after boasting the opposite was a 'for sure'. Again Show us your core duo example not business creds. Keep it related to civ4 interturn processing, not some 'Unreal' garbage

 

[Thanny]

Nah, I don't defend the choice I made. I defend what I see is true in these circumstances . I don't dismiss fact based on observation over another generalized speel on how much better core due architecture should make civ4 over top model monocore. You think I havn't heard it before. Thing is the prrof wasn't there to back it up but plenty was their to dismiss this as the most logical guestimate

1) Prescott is not the top model single-core processor. AMD still wins that one.

2) I have no doubt you've heard it before. But you clearly haven't learned that you're completely wrong.

Of course the Pent is dead, unless you just happen to be a civ player who chanced at the greatest mono perfomance for one third the price, then doubles it up nice with 64bit supported ram supply, 4.6ghz not breaking a sweat and plenty of L2 to make those turns fly.
You really think a 2.6 coreduo like Dano discribes is that much more efficient then a monocore blazing turns on mono designed software?

You still seem to think the high clock speeds of a P4 processor indicating blazing speed. They don't. They show how horribly designed those chips were, that they can't get anything done without running twice as fast as better chips.

And for the record, yes, the E6600 is still faster than a Prescott at a whopping 4.6GHz. Only just in some cases, but still faster. That's at stock speed, of course.

I know it hurts to hear an old computer beats you egostacks in some ways Heck, I heard the Intell marketing problem way back: Its convince buyers less is more, years after boasting the opposite was a 'for sure'. Again Show us your core duo example not business creds. Keep it related to civ4 interturn processing, not some 'Unreal' garbage

How about you provide some evidence for a change? I've already demonstrated beyond anything approaching a reasonable doubt that you're completely wrong. You've yet to provide anything but your assertion that a Prescott chip is faster for Civ4.

I doubt anything will be forthcoming, since that would require you take steps to stop denying reality.

The fact that I'm a programmer means I'm qualified to judge your claims about how Civ4 operates. That is, more qualified than I already am from having the game and noting how it runs.

And for the further record, I don't own a Core 2 processor. I have an Athlon X2 6000+ in this machine, which is the same speed as the E6600, though 3.0GHz instead of 2.4GHz. Still faster than a 4.6GHz Prescott running every bit of software, including Civ4. And eating one hell of a lot less power while doing it.

 

[Magic Cat]

Lots of conjecture & nonsense

Facts backed up with experience & evidence

As another Software Engineering professional, I'd like to give you some support Thanny. I fail to see how the Civ4 software could possibly have been written to run faster on an outdated processor architecture. That would be quite some feat.

Perhaps this is mono optimization?
if( !OldProcessor )
{
mainThread->Sleep( AGES );
}

 

[T.A JONES]

.

Perhaps this is mono optimization?
if( !OldProcessor )
{
mainThread->Sleep( AGES );
}

NO insider jokes. That could read klingon for all I now. JUST SHOW ME THE PROOF! Here I was thinking more along these lines:

LOCATION: Civ3 Civfantics modforum
EVENT: Far Horizons MEGA MOD MEGA MAP: Due vs Cedarmill
And ya plenty more to come I imagine. Wheres the benchmark you say thranny relates as evidence in coreduo's better civ4 utillzation over top pent monocore? I couldn't find it!

 

[Nansen]

Thanny has quoted it in his benchmark references on single thread performance.

There seems to be no one on this thread who supports the claim that CivIV is able to benefit from several threads or cores, so comparing with benchmarks for other single-threading software is a valid approach.
It also seems to be generally agreed that CivIV benefits more from additional memory and processor speed than graphics performance, so those are the key screws to turn to improve performance.

In addition (and this is not scientific at all), I run BtS both on a XP rig with a 3.0GHz P4 (multithreading), 1Gig RAM and an 6800 Graphics card on 1280x1024, and on a Notebook with a dual core T7200 at 2.0Ghz, 2 Gig Ram and a 7400Go graphics adapter under XP at 1280x800 (tried it yesterday also at 1280x1024) and the latter runs noticeably faster on standard and large maps. Haven't tried huge maps yet, games take too long for my taste anyway.
On the settings I run (Graphics quality high), the end game on a large map with about 8 AI civs never takes more than a few seconds between turns even on the single core rig.

Now you can discuss without end how much of that subjective difference in speed is related to the difference in memory and the more modern graphics adapter - that is why I only quote this as anecdotal evidence (and I am not even a professional programmer ).

 

[T.A JONES]

Good reads. I hope it will get better if my challenge IM about to present is accepted by a few levels of coredue. (sorry, for grammer. Please excuse the rushed job on my part for all prior posting )
Anyway back to it. I buy the best for what I need. I have obsession with fastest turns!! Thanny, other coredue engineers, programmers, need to admit its all speculative in this area. NO matter which way this turns out, the results will be very perswading to ether side and best, in my new CPU purchase and hopefully a few others to. (those who only use rig for net, work related or civ4 and majorty games made before and even after )

THINK HOW BIG THE WEB IS, HOW BIG CIV IS YET NO SPECIFIC BENCHMARK PROOF!!! . I count Over 20 other games that were picked. However Theres still plenty refences we can work with. We just have to find them here or make our own! (I bet lots would like to see me proved wrong. Well it will never happen! -unless you youtube your turn times! I will provide the map with the 1st video upload. Next I can send duplicates of my exact same save so we are all on the page.

Why not start with the civ3 comparison first? I figure no point making things over complicated having the subjective issues relating to differnces in ram or graphics all being thrown into the equation. I don't even have enough to ram at the moment to measure up to most you guys As of now I own 3 'inferior' Pentiums all for civ3. My best is the Pentium 4 codename: 3.2 ghz Cedarmill. ITs like the prescott I linked in every way just without 64 bit support thats better suited for civ4.

Still this will be the horse with my money riding on it till Ipick up the prescott. I will take on all challangers but remember its been claimed any core duo will outperform a Pentium and specifically the one I mentioned which was topclass, so a loss to my lesser model would be even more sad for coreduo's rep!

With civ3 we can truly get to the answer much faster. Its the Same mono designed principles as used in civ4 yet without 'contestants' having advantage due to varations in ram.
All civ3 needs to play any size map is 512mb. After 512ram ceilin is reached and your onboard g card is ditched, the only thing deciding as to your waits is your CPU.. We will use much larger then default "huge" for this demo.

Its perfect comparison to see truly whether or not this extra $1000 is making a differnce like you all say, or, like I say in many cases the upgrade is actually negative ( i believe negative will come from all the early string of coredue's )
IM sure you all agree this would be considered a valid approach?

 

[Thanny]

Anyway back to it. I buy the best for what I need. I have obsession with fastest turns!!
[snip]
Why not start with the civ3 comparison first? I figure no point making
[snip]

I don't have a functioning Civ3. My CD is scratched, so I can't re-install it (never played it much). Any test would have to be Civ4.

I suggest vanilla, and you provide the save game. Pick one which has the longest turn time from the save point, and say how long it took from pressing the turn button to the hourglass disappearing.

If I measure something slower, I'll say you're right, end of story. If not, I'll come up with an objective means of measuring turn time, along with a cheat-free method of reporting it, so we don't have to trust one another.

 

[T.A JONES]

I don't have a functioning Civ3. My CD is scratched, so I can't re-install it (never played it much). Any test would have to be Civ4.

I suggest vanilla, and you provide the save game. Pick one which has the longest turn time from the save point, and say how long it took from pressing the turn button to the hourglass disappearing.

If I measure something slower, I'll say you're right, end of story. If not, I'll come up with an objective means of measuring turn time, along with a cheat-free method of reporting it, so we don't have to trust one another.

You should trust me. I have nothing to gain. I took a side that surly would lose in any debate. I would never have taken it if i didn't come packing some series backup and most important, 1st hand xpierence to this crazy discovery.
It is simple to make the assumption same rules applys with civ4 once ram has been properly addressed. Thats why I so stressed the importance of the last stage in Pentium's Evo, Prescott's 64 bit support structure.
Because of this now Pentium is on the same playing feild as the civ3 benchmark where civ3 has been documented excelling (core due vs top model pentium)

Btw Corecell from MTI's has made overclocking so smooth the heat problem goes away. Just thought I'd mention that. IM sure a a few dimes on the powerbill can't really be worth more then the time you save if you an avid civ player
ANyway IM sure someone will take me up. Iv civ3 wins.( like It has in several linkable instances before) Then Id say IM on good ground to take it the civ4 stage where peoples coredue sad storys all over here have me convinced nothing will change. (I need only show this happend once and my aim is on all early model duos )

 

[Nansen]

Would be willing to participate in a test, besides the two computers already mentioned I could also offer a quad-core 2.4Ghz Q6600 with 3 Gig of Ram and an 8500GT graphics card (HP 9060n) for the upper end of the multi-core spectrum, but I have only Vista Premium (32bit) installed on that one.

I could probably hunt down my copy of Civ3 as well, although part of my stuff is in storage due to a recent move, but we would probably get a higher number of respondents if we use Civ4 Vanilla as Thanny proposed.

 

[T.A JONES]

Would be willing to participate in a test, besides the two computers already mentioned I could also offer a quad-core 2.4Ghz Q6600 with 3 Gig of Ram and an 8500GT graphics card (HP 9060n) for the upper end of the multi-core spectrum, but I have only Vista Premium (32bit) installed on that one.

I could probably hunt down my copy of Civ3 as well, although part of my stuff is in storage due to a recent move, but we would probably get a higher number of respondents if we use Civ4 Vanilla as Thanny proposed.

Right. In good time. Lets base the test on civ3. AS I mentiond early its out of stock due to holiday rush so Ive still to order the 64bit pentium that would put Pents at equals wth all you IMO.
Besides, what would civ4 benchtest prove that civ3 wouldn't? LIke one guy said, this is about avoiding confusing variables. If coreduo is losing in civ3 performace, your theory has got big problems. Lets drop this to lowest terms. Im tryin to provide the most concreate evidence for our efforts!

But hey THanks man!. Good your on board. I'll get crackin on a map for Conquests we can use as the 1st Web wide benchmark of the civ series!! If you havn't played civ3 in a while you'll be surprised in how fast turns go by in this new realm running off killer CPU's.

Heres sample of how civ3 looks on youtube!. Listen to the engine churn as we kick it into hyper drive, my 'lil' cedarmill an me ADD: (oh wait, thats not till next vid some 50 turns down the line-check my YT profle for link!)

Location: Far Horizons mod:

Event: AT the 1st point of surpassing 2 coreduos and countless other less powerful pentiums at a Huge map created max bottle neck point over 2/3's into the standard game. Both coredue crashed to infinty hang aka CTD (ref link in last post)

Game discription: Mod replicates 3rd xpak format. Xtra content includes more resources therfor higher volume of over-sea trade enabled during peak time, extra improvement calculations deriven during peak time, Expanded roster over 900 units, added wonders, sound effects expanded soundtrack que , all on 18 civ start count, massive map far greater then default and all the while operating high powered bluetooth 'ipoint' optical magnifier during course of AI turn calculatiions for optimium visualtion effects. 'wait time': under 30 seconds. AI warreport: 2 minutes

EXAMPLE OF CRAZY BREAK-NECK SPEEDS: Notice 6 boats drop their bombardments in the time it takes one in the old days.

 

[Thanny]

You should trust me. I have nothing to gain. I took a side that surly would lose in any debate. I would never have taken it if i didn't come packing some series backup and most important, 1st hand xpierence to this crazy discovery.
It is simple to make the assumption same rules applys with civ4 once ram has been properly addressed. Thats why I so stressed the importance of the last stage in Pentium's Evo, Prescott's 64 bit support structure.
Because of this now Pentium is on the same playing feild as the civ3 benchmark where civ3 has been documented excelling (core due vs top model pentium)

64-bit is irrelevant to Civ4. If you run it under 64-bit Windows, it will be using 32-bit emulation. Any machine with 1.5-2GB of RAM that's running 2K/XP is sufficient for the largest Civ4 game.

But I've given up on you. You keep writing things that don't make any sense, ignoring your difficulties with English. I'm not in the least convinced you actually know the difference between Prescott, Pentium D (which is basically two Prescott cores), Core Duo, and Core 2 Duo.

And you seem to be insisting on using Civ3 for tests, which I've already said I don't have access to.

So go ahead and keep playing in fantasy land, where your turns are zooming by on your old, egg-fryingly overclocked, poorly designed P4. The rest of us will get through more games in the same time on our cooler and more efficient modern chips.