Dolphin, the GameCube and Wii emulator - Forums

CPU: AMD Phenom II X4 955 3.2GHz
OS: Windows 7 64bit
Time: 17 minutes 34 seconds
Screenshot: http://i.imgur.com/5Xsl5d2.png

CPU: i5 3570k @ 4.2GHz (using motherboard's OC Genie 2 feature. Windows 8.1 inaccurately reports this as 5.13, as CPU-Z shows it's maxing out at 4.2)
OS: Windows 8.1 x64
Time: 10 minutes 0 seconds
Screenshot: http://i.imgur.com/3uHSFJj.png

No overclocked cpu's anymore. Delroth should really put that into the first post. :/

(01-21-2014, 11:38 PM)Anti-Ultimate Wrote: [ -> ]No overclocked cpu's anymore. Delroth should really put that into the first post. :/

Considering that he hasn't even added any results from the last 3 pages - results that have been there for 3 days now - I don't think that'll be happening anytime soon.

Basically, ever since the "no overclocking" rule got implemented, the chart hasn't been updated.

CPU: Intel Core 2 Duo E4500 @ 2.2GHz
OS: Windows 8.1
Time: 26 minutes, 55 seconds
Screenshot: http://i.imgur.com/251Bb7N.png

---

CPU: Intel Core i5 3230M @ 2.6GHz (Turbo Boost: 3.2GHz)
OS: Windows 8.1
Time: 16 minutes, 36 seconds
Screenshot: http://i.imgur.com/rmwlJDF.png

(01-21-2014, 11:38 PM)Anti-Ultimate Wrote: [ -> ]No overclocked cpu's anymore. Delroth should really put that into the first post. :/

Why?

For example the 3570k is an extremely popular cpu because you pay slightly more and you get a free, stable 400+mhz out of it right off the bat. More if you can do it.
Going forward in the future it's likely that even more cpus will be "unlocked" with the ability to have a stable clock at those frequencies. And what difference does it make to this benchmark really?

(01-22-2014, 01:47 AM)Gvaz Wrote: [ -> ]

(01-21-2014, 11:38 PM)Anti-Ultimate Wrote: [ -> ]No overclocked cpu's anymore. Delroth should really put that into the first post. :/

Why?

For example the 3570k is an extremely popular cpu because you pay slightly more and you get a free, stable 400+mhz out of it right off the bat. More if you can do it.
Going forward in the future it's likely that even more cpus will be "unlocked" with the ability to have a stable clock at those frequencies. And what difference does it make to this benchmark really?

According to delroth, OC is too complicated with Cache Ratio, BLCK changes and so on. He says he doesn't want to publish data taken from a CPU that is overclocked like ass only for this benchmark and will catch fire else. (He said it like this).

I think he should have said that first, or understood most people would overclock their cpu if given the choice and ability to do so. Also the prevalence of auto unlocked cpus since sandybridge

It is a bit stupid. If 90%+ of the userbase of a chip bought it specifically to overclock, because once it was overclocked it would be the fastest thing available, then it doesn't support the "the point is just to know how well a given CPU performs" idea. In Dolphin, it's a fact that users are going to find the 4670K faster than the 4670, because when they spend the extra money on the K, they're spending that money because they want to overclock.

There is a point to the idea that "overclocking adds so many dimensions to the problem that the results become meaningless", but most of the time, a stable overclock of a specific model of chip at a specific frequency will give pretty consistent performance between individual chips, and as the frequency is included in the results, it's not like people will muddle a 0.2 GHz OC with a 1 GHz OC.

Finally "people trying to optimize for a benchmark" isn't a huge issue, as if we're locked down to a specific revision, and a specific test which should be entirely CPU bound, the only things optimisation will bring about are overclocks and new CPU purchases, both of which should cause a direct increase in emulation speed in game, and are therefore relevant to Dolphin users (or prospective users) trying to decide what CPU to purchase or gauge their own performance before going to the hassle of ripping their game library.

I've probably said something dumb here which needs picking apart (and therefore should be), but in its current form, I don't see Delroth's reasoning making much sense.

Anyhow, I've got several older computers which I never got around to benching last time, but will bother to now the process is simpler.

---

(01-22-2014, 02:47 AM)Anti-Ultimate Wrote: [ -> ]

(01-22-2014, 01:47 AM)Gvaz Wrote: [ -> ]

(01-21-2014, 11:38 PM)Anti-Ultimate Wrote: [ -> ]No overclocked cpu's anymore. Delroth should really put that into the first post. :/

Why?

For example the 3570k is an extremely popular cpu because you pay slightly more and you get a free, stable 400+mhz out of it right off the bat. More if you can do it.
Going forward in the future it's likely that even more cpus will be "unlocked" with the ability to have a stable clock at those frequencies. And what difference does it make to this benchmark really?

According to delroth, OC is too complicated with Cache Ratio, BLCK changes and so on. He says he doesn't want to publish data taken from a CPU that is overclocked like ass only for this benchmark and will catch fire else. (He said it like this).

That makes a little more sense than how I read it originally. Surely the ideal solution would be to request only stable overclocks, and just hope that no-one lies. If people were going to lie about that, they could also have 'shopped the numbers in the screenshot. If someone cares about winning enough to burn their CPU out, then there are cheaper and easier ways to generate useless results.

I got myself a good chip that would go to 4.6Ghz at like 1.25V. Nothing really bad right here.

Now, there are also people trying to beat my 4.6Ghz score with like 1.4-1.5V Suicide runs. This is bad.