Those results seem to support the concepts put forth by Stilt in that other website discussion, where different ranges of memory speeds can trigger the card to use different memory latencies. So if you change the memory speed a little, and stay in the range for one memory latency, the hash rate doesn't change much. But, if you change the memory speed a little, and happen to hit a range that uses a different memory latency, then you see the difference in your hash rate. Even if you use a slower memory speed, this can happen, where the card will just happen to use better latencies for that slower memory speed, and boom the hashrate goes up. Ideally you just fix the table of latencies used at the various frequencies, but the manufacturers were usually too sloppy to bother, so the latency values are all over the place.
ON a side note, I wanted to encourage the healthy discussion regarding mining vs buying, and mining litecoins vs other alt coins. I think Blast and Silver are doing a great service by being champions of these discussions. Ultimately both positions help all miners. Think about it, if people switch to mining alt coins to speculate on various different currencies, the difficulty is spread out among lots of currencies and gives more opportunity for all to mine without dogpiling onto one single currency where the difficulty would then be blown out of the water. And for mining vs. buying, think about it, if there are more buyers of various coins, that just drives up the price for those who are mining. So it is all helpful and please keep discussion going, and even argue about it passionately guys!
Thanks, I've had people post and PM me thanking me for analysis as well, but I want to note that my discussion of mining arbitrage was 100% about mining altcoins vs litecoins. There's so much randomness in life that it's hard to separate skill from luck without looking at much longer periods of time than just a month or two. Unless there's a super-outlier like dogecoin that offered such a gigantic, huge, and unprecedented arbitrage gap making the risk virtually zero to arbitrage it, it is often difficult to predict what's best to mine thanks to continually shifting prices, difficulties, and introductions of new coins. It's definitely possible to "beat the market" (i.e., make more by mining an altcoin than litecoin) but it's also possible to underperform the market as well, by mining something that turned out to be a less profitable than you expected. This is especially true for altcoins that aren't on exchanges or are on exchanges with limited market depth and small market cap since prices and difficulty can be extremely volatile. Basically, other than weird times like dogecoin, you cannot claim that "oh, I can stop mining altcoin at any time and pick it up again when it's more profitable." That's b.s., because you don't actually know your profitability until AFTER you've already decided what to mine. It's only after you mine and sell and pay exchange fees that you really know if you beat the market or not. What Silver's condescending posts claimed was basically: if you can predict the future, you can beat the market. Well DUH.
Many people try to time the market, think it's free money and they can buy low and sell high. Many people fail because it's freaking hard to time the market. When bitcoin crashed down to something like $500 recently, how many people correctly timed the market and rode it back up? Right up to the weekend before, Lehman Bros. was still in business. Right up to the point where Bitcoin Savings and Trust failed, there were people who KNEW it was a ponzi scheme but told themselves that they would somehow "know" when it was time to bail, so self-assured they were. People were still trading on MF Global right up to the point of failure. Etc. Yet here we have a guy Silverforce who claims he can simply jump in and out of various altcoins to maximize his profitability, or that he can turn the job over to a pool operator who can do the same thing. Sure, if competition for arbitrage opportunities is low enough, you can definitely beat the market, even after subtracting hefty coinhopping pool fees, but there is a LOT of of competition for arbitrage opportunities these days, making it quite difficult to consistently beat the market (profit more than mining litecoin, after accounting for pool fees and exchange fees) for long periods of time (at LEAST several years). Short lucky periods don't mean squat. If he has those kinds of market-timing skills he should work in finance and make a killing.
Mining straight LTC at a 0% fee pool with good uptime (e.g., WeMineLTC) is like buying a S&P500 Index Fund. It's boring but it's low-fee and less volatile than attempting to pick your own stocks. Attempting to arbitrage mining profitability directly or via a coinhopping pool is like trying to pick stocks yourself or outsourcing it to someone else via a high-fee mutual fund (this is an imperfect analogy and mining altcoins is possibly not quite as bad as this because of very rare coins like dogecoin that allow for brief periods of relatively low-risk arbitrage, but on the other hand, money managers can take advantage of arbitrage situations as well in finance).
Very few people manage to beat the market by picking stocks, over long periods of time. Short periods of time is easy, you can simply luck your way into beating the market for a long time, even years. Longer term? Ehhh. Historically there have been very few mutual funds that have consistently beaten the S&P500 (or some other index with hundreds or thousands of companies listed that is supposed to be representative of all stocks), especially once you add back in all the failed funds that dissolved and all their negative returns. (If you look only at the mutual funds that still exist today, that's cherrypicking data, because only the more profitable mutual funds survive.) Due to sheer luck, there are some funds that beat the S&P500 for several years in a row, and some funds that underperform the S&P500 for several years in a row if they don't get shut down first. It's only when a fund outperforms the S&P500 for decades that you can be reasonably certain that there is skill involved and not just luck (e.g., it's statistically very unlikely that Warren Buffett is merely lucky; it's highly probable he is skillful, too). That's why you should never trust conmen trying to pump and dump penny stocks based on a brief period of time. Even during the unique dogecoin run, nobody knew for sure when the gravy train would run out. Hindsight is 20/20. Timing altcoins is arguably EVEN HARDER than timing the stock market.
Altcoins are basically the penny stocks of the cryptocurrency world, so when you mine and hold them, however briefly, you expose yourself to the altcoin markets. For every guy loudly bragging about how they made money on penny stock X (sometimes as part of a pump-and-dump scheme), there are silent guys who lost on penny stock Y. Penny stocks are highly volatile, low-market cap, low-liquidity, and much more easily price-manipulated than large-cap. What we see in cryptocurrency is what we used to see more of in real-life penny stocks: market manipulation, insider trading if you know the altcoin developers' plans, pump and dump schemes, embezzling pool operators, pool and exchange operators that shut down claiming they were hacked but may have simply absconded with customer funds, etc. Unlike real-life stocks, there are no SEC laws to help keep a lid on fraud.
As always, don't gamble more than you can afford, and good luck.