239 replies to this topic

[Oderint dum Metuant]

Someone want to point out the better per core performance of the old Phenom series chips over FX? (or at least certainly better than Faildozer)

[Odins Fist]

SappInfernaWulf, on 30 January 2015 - 07:31 AM, said:

First off the Intels do give a much better reaction to FPS. However its only on the higher end AMD chips that seem to fall on their face.

AMD the only chips we have found that actually work great with MWO,

EVGA GTX 750 TI, beast


This is my MWO RIg, I run a MSI AM3+ 970A-G43 Mobo a AMD Athlon II x4 640, 8 gb of Ripjaw X PC3-10666, a Samsung 840 500 gb SSD. and a WD Red WD40EFRX 4TB drive. and last but not least a incredible Palit Geforce GTX 560 TI, Im running MWO at 58-60 FPS on pug drops with only a slight dip in CW performance, framerates are stable at 45-51 in CW. All these stats mind you are at max settings DX 11@ 1920x1080 res, so all you who say it cant be done with OLD PARTS need to back off cause it can be done.

AGHHH.. HA HA HA HA HA HA HA HA HA HA HA... ROFLMAO.... HA HA HA HA HA HA HA HA HA...
GASP..... HA HA HA HA HA HA HA HA HA HA HA HA HA..

JA, I remember my first 560 ti..!!!
.

.
Pulled better frames with my Phenom IIx6 1100t and a GTX 560 ti then a buddy with an ATI 6850 & FX-8350 that was clocked 400 mhz higher then me on CERTAIN settings in MWO.

AMD had a FLOP with the FX-series CPUs, they have even admitted that publicly.

Saying ANYTHING can beat an AMD FX-series CPUs in certain tasks at CLOCK versus same CLOCK is not really saying much.

Those of us that say how much a fail the FX-series CPUs were BEFORE buying one, never bought one.

FYI, I was GIFTED a FX-8 core CPU, and I NEVER installed it on either of my Top End Asus 990-FX mobos, I figured if the Thubans I have in them ever die, then I will use it, BUT not until then.

EDIT: Not to get funny, BUT some lines in your post seemed almost contradictory.

Edited by Odins Fist, 30 January 2015 - 12:16 PM.

[Odins Fist]

DV McKenna, on 30 January 2015 - 09:20 AM, said:

Someone want to point out the better per core performance of the old Phenom series chips over FX? (or at least certainly better than Faildozer)

I have been for literally YEARS on the Hardware Forums in MWO.
Literally years.

Edit: Single Threaded apps

Edited by Odins Fist, 30 January 2015 - 10:31 AM.

[Iqfish]

Odins Fist, on 30 January 2015 - 09:35 AM, said:

snip

"maybe if I write in caps, people will get my point"

[Odins Fist]

Iqfish, on 30 January 2015 - 09:54 AM, said:

"maybe if I write in caps, people will get my point"

STFU...

So do you have anything relevant to add to this thread..??

No..?? Ok then GTFO..

CAPS got your attention didn't they, now run along... Shooo, go on.

Edited by Odins Fist, 30 January 2015 - 09:58 AM.

[Kuritaclan]

xWiredx, on 30 January 2015 - 06:06 AM, said:

Then all 200 of your samples are subpar. Lots of Haswell-E chips here hitting 4.5GhZ and higher on less than 1.35v, and at 1.35v people are getting 4.5GhZ and higher. (btw, 1.35v I believe Intel has posted as the official 'safe' limit, I can't remember).

http://www.overclock...ard-owners-club

As far as DDR4-goes, MWO doesn't care. MWO doesn't care after DDR3-1866 at all, and that is only a very tiny improvement over 1600 (like... 1-3fps max). We've been over this a few times already. The major technical advantage of DDR4 is massive bandwidth increase using less voltage. To really see it leap ahead, you are exactly correct in thinking that DDR4 needs to mature a lot still. Once DDR4-3000 and faster becomes mainstream, DDR4 will be more worthwhile.

Look in the Leaderboard you posted. There are out of all Chips only 4Chips of 5960X that hitting the the circumstances >4,5Ghz with <1,35V.And yes many samples are subpar. No there arn't much Haswell-E 8 Cores which hit the requirments. I Didn't mentioned 6 core Haswell-Es at all. But anyway to get them on high clockes you need a customized water loop and this can get pretty expensiv additional to the cpu itself and the mainboard and the new RAM (so in the end you will gain some FPS for mostlikely double the price). Intel never sayed anything about the safe limit. Since this is only written in the VDR and this paper is under NDA for Haswell and Haswell-E.

Hope so too with DDR4, maybe Skylake has some Modules which are on same level with Latency.

[xWiredx]

With the latest Intel roadmap leak, another positive to going on X99 has come up: Broadwell-E will likely be compatible with current X99 boards. This is certainly something I'm interested in, and anybody that was held up on whether to go X99 or not based on future life can now buy a little more confidently.

The whole point of posting the overclock leaderboard was to show that with less than 50 samples, several met the criteria. If you had 200 and not a single one was capable of doing it, one of the data sets is a statistical anomaly. I'd be betting on your set being the anomaly since the leaderboard featured all 3 models of chip and from different fabs. It's a pretty good mix.

Most people aren't interested in the 5960X, most are interested in the 5820K and 5930K when it comes to gaming. My system is at 1.27v and 4.37GhZ (it isn't on that leaderboard, I'm not a member of their forum). It has some breathing room still. The H105 I have cooling it has not let a core temp go past 54c yet. I like my current speed, though, don't really need to push it further.

[Kuritaclan]

You just don't get it. The Leaderboard is in itself no criteria. Since these high numbers are done mostlikely with selceted CPUs! You just be false minded if you can use every cpu to chase for high core clocks. - As i said it is not my sample it is the sample size of 200 CPUs to meet this requirment for further overcloking on the edge. It is as hard to get a i7 5960X with >4,5Ghz and <1,35V as it is to get a i7 4790k with 5,0Ghz and <1,25V or in the case if a i7 5820k with >4.7GHz @ <1,3V.

Edited by Kuritaclan, 30 January 2015 - 07:16 PM.

[Catamount]

Kurita has a point. Every entry onto that leader board is an anecdote, not part of a random sample. All that's required for that leader board to be full is for there to be enough chips to make it on there... in the whole world. Obviously if you can hand pick the best chips on the planet, or at least hand pick the best chips out of all the people, everywhere, who might want to post up bragging numbers, then you're going to get enough good numbers to fill one forum leaderboard.

When those good numbers don't show up in any of a sample of 200 CPUs, that leaves pretty low odds that you, personally, or me, or Kurita, or any given person, will happen to be someone who actually gets one of those good CPUs. Of course, if the 5960X is particularly difficult to get very high clocked then the odds will of course be somewhat better with 5820s and 5930s. How much? We'd need another random sample

Edited by Catamount, 30 January 2015 - 09:32 PM.

[xWiredx]

The leaderboard is not made of just the best chips (those would be anomalies). It is made of all chips with statistics submitted to it. Chips are from different areas, from different retailers, from different fabs, used by different overclockers with different cooling systems on different motherboards. That is about as random of a sampling as you can get. Considering more than half of the results are from day one chips, too, there is no way that the chips are hand-picked.

Also, calling it an anecdote is relatively incorrect as these are all validated numbers. Where else would you get proper statistics than a database full of numbers (in fact, the exact numbers that have been submitted and validated on the above-linked leaderboard). I mean, really, did we fail statistics analysis class or something?

[Smokeyjedi]

xWiredx, on 31 January 2015 - 06:13 AM, said:

The leaderboard is not made of just the best chips (those would be anomalies). It is made of all chips with statistics submitted to it. Chips are from different areas, from different retailers, from different fabs, used by different overclockers with different cooling systems on different motherboards. That is about as random of a sampling as you can get. Considering more than half of the results are from day one chips, too, there is no way that the chips are hand-picked.

Also, calling it an anecdote is relatively incorrect as these are all validated numbers. Where else would you get proper statistics than a database full of numbers (in fact, the exact numbers that have been submitted and validated on the above-linked leaderboard). I mean, really, did we fail statistics analysis class or something?

I have to agree with most of this, Think of how many people have bought 5960s or 5820s and have no idea/notion of how to OC, who will never ever see that graph chart or leader board who are sitting on those exact bin worthy chips........way more than 50% so that chart isnt say all end all of this conversation.

average non intuative PC use > tech saavy OC guru willing to submit data for said charts**

there are sooo many samples unaccounted for in the world, and soo many variables.
*geolocation/climate/season/humidity/ambient air quality, ETC..*

I damwell know a dude running custom loops in northern canada is going to get better temps than exact rig in Texas with same exact loop, Its all relative.

Edited by Smokeyjedi, 31 January 2015 - 06:52 AM.

[conquistadorst]

Don't be tempted by the cost savings. You will regret the performance and spend your time convincing yourself it's not the CPU when it is. The last AMD CPU I used was the last good one they made, when they first came out with their quad core CPUs. That was it, never again have they restored their former glory.

[Bill Lumbar]

conquistadorst, on 31 January 2015 - 07:31 AM, said:

Don't be tempted by the cost savings. You will regret the performance and spend your time convincing yourself it's not the CPU when it is. The last AMD CPU I used was the last good one they made, when they first came out with their quad core CPUs. That was it, never again have they restored their former glory.

The very first Quads they came out with kinda blew compared to Intel's offerings at the time, (Phenom's) The Phenom II's are what I believe you meant?

[Catamount]

xWiredx, on 31 January 2015 - 06:13 AM, said:

The leaderboard is not made of just the best chips (those would be anomalies). It is made of all chips with statistics submitted to it. Chips are from different areas, from different retailers, from different fabs, used by different overclockers with different cooling systems on different motherboards. That is about as random of a sampling as you can get. Considering more than half of the results are from day one chips, too, there is no way that the chips are hand-picked.

Also, calling it an anecdote is relatively incorrect as these are all validated numbers. Where else would you get proper statistics than a database full of numbers (in fact, the exact numbers that have been submitted and validated on the above-linked leaderboard). I mean, really, did we fail statistics analysis class or something?

My, aren't we the touchy one?

Anecdote doesn't mean information wasn't validated. I can validate an anecdote in any way I want and it's still an anecdote. The reason something is anecdotal is because it's data without statistical context. I could easily point to sixteen parents who's children all got diagnosed with autism after an MMR shot, but that's not terribly useful information. Are those sixteen the only people out of everyone? I don't know. How big is the population from which they're drawn? From the information provided, I don't know. Consequently, how big a random sample would be required to be likely to get another 16 people? Once again, I don't know. I know nothing to make a statistical inference from these sixteen people, presented in a vacuum. Being individual cases in a vacuum of information is what makes it anecdata. It has absolutely, positively, nothing to do with validation.

I could make a quip about failing statistics class right now, but I'll leave that stuff to you.



Now, I'm looking at this leaderboard, and all I see are sixteen numbers in a vacuum. Are those numbers typical? I don't know. They're selectively submitted from the total population, and I have absolutely, positively, no basis to make a statistical inference about how typical those results are. They might be entirely typical, or extreme outliers. Without the information to make such statistical inferences, your leaderboard is anecdata, sorry.

Did I miss something in the page you posted? If so, then instead of non-useful snark, maybe you could actually point it out.



Edit: HERPDERP my damn browser wasn't scrolling the list.

Okay, so it's 113 entries, but the same statistical problems remain. We didn't pick 113 people at random out of the population of Haswell-E owners and say "tell us your overclock", these are chips listed precisely because they're the best entries to the site. That is the definition of a leaderboard, is it not? It doesn't matter if the leadboard is picked from everyone. Even if it's only picked from a small subset of everyone, it's still hand-picking from that subset.

Does that mean they're non-typical results? No. It means we have no idea how typical they are. Different test conditions, ie a semi-random sample of conditions, does not mean random sample of chips.

This might be entirely typical, or entirely non-typical, and there is absolutely no way to tell from 113 instances of people coming forward with good OCs in a vacuum, because we didn't pick those 113 people before we knew their OCs and try to get some kind of distribution on their OC ability. In other words, it's still anecdata.

Edited by Catamount, 31 January 2015 - 01:40 PM.

[Kuritaclan]

Catamount, on 31 January 2015 - 10:44 AM, said:

Okay, so it's 113 entries, but the same statistical problems remain. We didn't pick 113 people at random out of the population of Haswell-E owners and say "tell us your overclock", these are chips listed precisely because they're the best entries to the site. That is the definition of a leaderboard, is it not? It doesn't matter if the leadboard is picked from everyone. Even if it's only picked from a small subset of everyone, it's still hand-picking from that subset.

Lets name it. The Leaderboard is an indicator. You may reach with a cpu out of the box such numbers or not. Some of the enterys are from preople who oc and got lucky so they show their numbers. Others who are in the list have the chance by owning a pc-shop and select the cpus.

At least for now in the matter of gaming a 6core or 8 core (Haswell-E) cpu is not worth the price. Maybee in 3 or 5 years this will be the case. Or at least with 6 core skylake consumer plattform it would be reasonable that more optimization could come in.

[Catamount]

I think we might well be at a point where the 5820k is worth pursuing for some gaming rigs. It's not that expensive, really, and MWO really does love it.

Nevertheless, your mileage may vary, and we don't know by how much, because we lack any solid data on what a typical Haswell-E can overclock to. I still have no qualms about recommending them if the budget is there.

[Smokeyjedi]

Catamount, on 31 January 2015 - 09:14 PM, said:

I think we might well be at a point where the 5820k is worth pursuing for some gaming rigs. It's not that expensive, really, and MWO really does love it.

Nevertheless, your mileage may vary, and we don't know by how much, because we lack any solid data on what a typical Haswell-E can overclock to. I still have no qualms about recommending them if the budget is there.

An no, A few IPC a tonne of memory bandwidth and some extra cores? who would want that for gaming? derp/

[-Wulf-]

cSand, on 11 January 2015 - 09:35 AM, said:

Hey I just thought I'd chime in here.

If you are on a budget, and don't want to OC

Then go with a lga 1150 XEON.

You can't overclock it, but it's an i7 for an i5 price. Difference is no iGPU but big deal, since you are having a dedicated graphics card.

Got my Xeon v3 1231 (3.4-3.8GHz) for 270$ CDN and haven't looked back. MWO runs awesomely.

Some people will tell you they are only for servers or whatever but they have no idea what they are talking about. It is literally in i7 processor, selected for stability, with no Intel onboard graphics.

You can probbaly even pick up a 1241 v3 (3.5-3.9GHz) for cheap these days as the new lines are coming out soon.

Seriously man, don't go i5 if you don't care for overclocking. Go Xeon like a champ and buy smart.


As for AMD there is the 8370 now which is the "new" one but it's just the same old FZX chip clocked higher. That said, my friend, and my brother, both own an d8320 and 8350 respectively and have no issues with MWO at all (bro has a 760 and runs this game maxed out with no hiccups). So take what you will from that. AMD's APUs are quite good but you can see that they're going in that direction and away from their old ways of just balls deep power. The FX chip is a few years old now and hasn't really been updated.

Anyways.

The cSand verdict: go Xeon don't be a chump

Edit: additional bonus, is you can go with a H97 based motherboard and not have any wasted potential, and save some $$

AGREED CSAND on the Xeon, However I run a I5-2500k, something that I have found out is the stability of the I5-2500k series and the I7-2700k both are incredible for thier thermals. The 2X00 series I7 and I5's can be taken to 4.8 ghz stable on a standard air cooling system. For all you non over clockers out there go for the Xeon and to hell with the detractors saying they are erver chips only, the wife uses a server HD for the 128mb cache on her 4 tb HDD. People say it dont work like that but cant argue when she posts screens of her system running it so be mindful of the fact that server parts can really provide the pickmeup that some gaming rigs need.

Also AMD makes great stuff however for MWO only 2 AMD chips seem to shine. The Phenom II x4 945 Black, is the weaker of the two I have found its a great processor but needs to be OC'd 400 mhz over at 3.8ghz to get the 60 fps we use as a pass fail benchmark. However the AMD BEAST is the Athlon II x4 640, this chips is just stupid sick, this AMD is an old chip but DAYUM It literally kicks the sheep spit out of any newer AMD processor out there, at least in mwo. it dont require any over clocking and can just chug right through MWO, you need a great gfx card to put with it but when you do this processor can run MWO at 60 FPS with no hiccups. We run our Athlon II x4 640 with a 2 gb Palit GeForce GTX 560TI. all in all most people think that old parts or server parts cant be used in a gaming rig, its just not true. Sometime the older equipment can really suprise you.

[Smokeyjedi]

Falkwulf, on 01 February 2015 - 06:52 AM, said:

AGREED CSAND on the Xeon, However I run a I5-2500k, something that I have found out is the stability of the I5-2500k series and the I7-2700k both are incredible for thier thermals. The 2X00 series I7 and I5's can be taken to 4.8 ghz stable on a standard air cooling system. For all you non over clockers out there go for the Xeon and to hell with the detractors saying they are erver chips only, the wife uses a server HD for the 128mb cache on her 4 tb HDD. People say it dont work like that but cant argue when she posts screens of her system running it so be mindful of the fact that server parts can really provide the pickmeup that some gaming rigs need.

Also AMD makes great stuff however for MWO only 2 AMD chips seem to shine. The Phenom II x4 945 Black, is the weaker of the two I have found its a great processor but needs to be OC'd 400 mhz over at 3.8ghz to get the 60 fps we use as a pass fail benchmark. However the AMD BEAST is the Athlon II x4 640, this chips is just stupid sick, this AMD is an old chip but DAYUM It literally kicks the sheep spit out of any newer AMD processor out there, at least in mwo. it dont require any over clocking and can just chug right through MWO, you need a great gfx card to put with it but when you do this processor can run MWO at 60 FPS with no hiccups. We run our Athlon II x4 640 with a 2 gb Palit GeForce GTX 560TI. all in all most people think that old parts or server parts cant be used in a gaming rig, its just not true. Sometime the older equipment can really suprise you.

http://www.cpubenchmark.net/compare.php?cmp[]=171&cmp[]=1910
The single thread performance of the FX is 50% higher according to this...........than the Athlon 2 X4 640................I couldnt possibly understand how that could work for you........even of you dropped to 3 of 6 threads on the Vishera @ 3.9ghz technically that would still match IPC/Compute power.......not to mention the lack of memory bandwidth that limited DDR3 speeds carry.

Are you playing in 1920X1080P? What am I missing??

[Goose]

Log files or it didn't happen …